New Chemistry and New Opportunities from the Expanding Protein Universe Universe Protein Expanding the from Opportunities New and New Chemistry

Copyright of the works in this eBook is vested with World Scientific Publishing. The following eBook is allowed for host on http://www.solvayinstitutes.be/ only and may not be resold, copied, further disseminated, or hosted on any other third party website or repository without the copyright holders' written permission. http://www.worldscientific.com/worldscibooks/10.1142/9180

For any queries, please contact [email protected].

distribution is allowed. distribution New Chemistry and New Opportunities from the 9180

only. No further No only. Expanding Protein Universe

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

July 25, 2013 17:28 WSPC - Proceedings Trim Size: 9.75in x 6.5in icmp12-master

This page intentionally left blank distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

23rd International Solvay Conference on Chemistry

New Chemistry and

New Opportunities from the Expanding Protein Universe distribution is allowed. distribution Hotel Métropole, Brussels 16 – 19 October 2013

9180 only. No further No only.

Scientific Editors

vayinstitutes.be/ Kurt Wüthrich PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- The Scripps Research Institute, USA & ETH Zürich, Switzerland Ian A. Wilson http://www.sol The Scripps Research Institute, USA

host on on host Donald Hilvert ETH Zürich, Switzerland Dennis W. Wolan http://www.worldscientific.com/worldscibooks/10.1142/ The Scripps Research Institute, USA

International Solvay Institutes Editor Anne De Wit ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Université Libre de Bruxelles, Belgium NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

World Scientific

NEW JERSEY • LONDON • SINGAPORE • BEIJING • SHANGHAI • HONG KONG • TAIPEI • CHENNAI Published by World Scientific Publishing Co. Pte. Ltd. 5 Toh Tuck Link, Singapore 596224 USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601 UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

Library of Congress Cataloging-in-Publication Data International Solvay Conference on Chemistry (23rd : 2013 : Brussels, Belgium) New chemistry and new opportunities from the expanding protein universe : proceedings of the 23rd International Solvay Conference on Chemistry, Hotel Métropole, Brussels 16–19 October 2013 / editor, Kurt Wüthrich, The Scripps Research Institute, USA & ETH Zürich, Switzerl [and 4 others]. pages cm distribution is allowed. distribution Includes bibliographical references and index.

ISBN 978-9814603829 (hardcover : alk. paper) -- ISBN 9814603821 (hardcover : alk. paper) -- ISBN 978-9814616560 (pbk. : alk. paper) -- ISBN 9814616567 (pbk. : alk. paper) 9180 1. Proteins--Synthesis--Congresses. 2. Enzymes--Congresses. I. Wüthrich, Kurt, editor. II. Title. QP551.I5188 2013 only. No further No only. 612'.01575--dc23 2014024092

British Library Cataloguing-in-Publication Data vayinstitutes.be/ A catalogue record for this book is available from the British Library. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol

host on on host Copyright © 2014 by World Scientific Publishing Co. Pte. Ltd. All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any information storage and retrieval system now known or to be invented, without written permission from the publisher. http://www.worldscientific.com/worldscibooks/10.1142/

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Printed in Singapore October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg v

Contents

I. The International Solvay Institutes xi II. Solvay Scientiﬁc Committee for Chemistry xiii distribution is allowed. distribution III. Acknowledgements xv

IV. Participants xvi 9180 V. Auditors xviii only. No further No only.

VI. Opening Address by Professor M. Henneaux xix VII. Preface by Professor K. W¨uthrich xxi

Session 1: New Chemistry in the Expanding Protein Universe 1

vayinstitutes.be/ 1.1 Novel Chemistry Still to be Found in Nature 3 PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- C. T. Walsh

1.2 Natural Product Biosynthesis in the Genomic Age 6 http://www.sol W. A. van der Donk host on on host 1.3 Peptide Dendrimers and Polycyclic Peptides 11 J.-L. Reymond

1.4 What can Comparative Genomics Reveal about the http://www.worldscientific.com/worldscibooks/10.1142/ Mechanisms of Protein Function Evolution? 16 N. L. Dawson, R. Studer, N. Furnham, D. Lees, S. Das, J. Thornton and C. Orengo

1.5 Exploring Chromatin Biology Using Protein Chemistry 20 T. W. Muir ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 1.6 Our Expanding Protein Universe 23 NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY A. Godzik

1.7 The Scientiﬁc Impact of Freely Available Chemical Probes 28

A. M. Edwards

1.8 Discussions of Session 1 31 October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg vi

vi Contents

Session 2: Exploring Enzyme Families and Enzyme Catalysis 57 2.1 Mechanistic Enzymology and Catalyst Design 59 D. Hilvert

2.2 Looking in New Directions for the Origins of Enzymatic Rate Accelerations 64 J. P. Klinman

2.3 Computational Enzyme Design and Methods to Predict the Role of Remote Mutations 70 distribution is allowed. distribution

K. N. Houk

9180 2.4 Discovering Novel Enzymes, Metabolites and Pathways 75 J. A. Gerlt only. No further No only.

2.5 Programming New Chemistry into the Genetic Code of Cells and Animals 79 J. W. Chin vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 2.6 Expanding the Enzyme Universe through a Marriage of Chemistry and Evolution 84

http://www.sol F. H. Arnold

2.7 Controlled Radical Reactions in Biology and the Importance of host on on host Metallo-Cofactor Biosynthesis 90 J. Stubbe

2.8 Discussions of Session 2 94 http://www.worldscientific.com/worldscibooks/10.1142/

Session 3: Microbiomes and Carbohydrate Chemistry 117 3.1 Structural Basis for Host/Commensal-Microbe Interactions in the Human Distal Gut Microbiome 119 I. A. Wilson

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 3.2 Carbohydrate Chemistry and Biology 125

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY C.-H. Wong

3.3 Chemical Biological Proteomics of Bacterial Protein Functionalities in the Human Distal Gut Microbiome 130 D. W. Wolan October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg vii

Contents vii

3.4 Automated Oligosaccharide Synthesis: From Insights into Fundamental Glycobiology to Vaccines and Diagnostics 136 P. H. Seeberger

3.5 Carbohydrate-Active Enzymes in Microbiomes 141 B. Henrissat

3.6 The Microbiome(s): Microbiota, Families, Functions 146 A. Godzik distribution is allowed. distribution 3.7 N-Linked Protein Glycosylation 150

M. Aebi 9180 3.8 Discussions of Session 3 154 only. No further No only.

Session 4: GPCRs and Transporters: Ligands, Cofactors, Drug Development 177 4.1 GPCRs and Transporters: Ligands, Cofactors, Drug Development 179 vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS G. von Heijne

4.2 Studies of GPCR Conformations in Non-Crystalline Milieus 182

http://www.sol K. W¨uthrich

host on on host 4.3 The Seven Transmembrane Superfamily 186 R. C. Stevens

4.4 Nanobodies for the Structural and Functional Investigation of GPCR Transmembrane Signaling 192 http://www.worldscientific.com/worldscibooks/10.1142/ E. Pardon and J. Steyaert

4.5 The Hidden Pharmacology of the Human GPCR-ome 195 B. L. Roth

4.6 Structures and Reaction Mechanisms of ABC Transporters 199

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World K. Locher

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 4.7 Discussions of Session 4 203

Session 5: Biologicals and Biosimilars 229 5.1 Biologicals and Biosimilars 231 S. Ghose and M. G. Gr¨utter October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg viii

viii Contents

5.2 Platform Technologies for the Artiﬁcial Pseudo-Natural Product Discovery 237 H. Suga

R R

5.3 Anticalins & Pasylation : New Concepts for Biopharmaceutical Drug Development from Protein Design 243 A. Skerra

5.4 From Natural Antibodies to Synthetic Proteins 249 S. S. Sidhu distribution is allowed. distribution

5.5 From Intact Antibodies to Armed Antibodies 254 D. Neri 9180 only. No further No only. 5.6 Regulating Cellular Life Death and Development Using Intracellular Combinatorial Antibody Libraries 257 R. Lerner, J. Xie, H. Zhang, K. Yea, J. Blanchard and K. Baldwin vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 5.7 Nanobodies: A Universe of Variable Domains and a Toolbox for Many Trades 260 L. Wyns http://www.sol 5.8 Discussions of Session 5 266 host on on host

Session 6: Proteins in Supramolecular Machines 285 6.1 Assembly of Filamentous Type 1 Pili from Uropathogenic Escherichia Coli Strains 287 http://www.worldscientific.com/worldscibooks/10.1142/ R. Glockshuber

6.2 HIV Envelope and Inﬂuenza Hemagglutinin Fusion Glycoproteins and the Quest for a Universal Vaccine 293 I. A. Wilson

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 6.3 Deconstruction of Iterative Polyketide Synthases 299 C. A. Townsend NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

6.4 Regulating Ribosome Pausing During Translation 305

M. V. Rodnina

6.5 The Molecular Mechanics of the Ribosome 308 J.Zhou,L.Lancaster,Z.Guo,J.P.DonohueandH.F.Noller October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg ix

Contents ix

6.6 Exploring the Dynamics of Supramolecular Machines with Cryo-Electron Microscopy 313 J. Frank

6.7 Crystallographic Studies of Eukaryotic Ribosomes and Functional Insights 318 N. Ban

6.8 Discussions of Session 6 323 distribution is allowed. distribution Index 345

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

July 25, 2013 17:28 WSPC - Proceedings Trim Size: 9.75in x 6.5in icmp12-master

This page intentionally left blank distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg xi

The International Solvay Institutes

Board of Directors distribution is allowed. distribution Members Mr. Jean-Marie Solvay

9180 President only. No further No only. Professor Franz Bingen Vice-president and Emeritus - Professor VUB Professor Benjamin Van Camp Secretary - Former Rector VUB vayinstitutes.be/ Chairman of the Board of Directors UZ Brussel PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Mr. Philippe Busquin Minister of State http://www.sol Professor Eric De Keuleneer

host on on host Solvay Brussels School of Economics & Management Mr. Bernard de Laguiche Chief Financial Oﬃcer Solvay S.A. Mr. Alain Delchambre http://www.worldscientific.com/worldscibooks/10.1142/ Chairman of the Board of Directors of the ULB Baron Daniel Janssen Honorary Chairman of the Board of Directors of Solvay S.A. Mr. Eddy Van Gelder Chairman of the Board of Directors of the VUB ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg xii

xii

Honorary Members Baron Andr´e Jaumotte Honorary Director Honorary Rector and Honorary President ULB

Mr. Jean-Marie Piret Emeritus Attorney General of the Supreme Court of Appeal and Honorary Principal Private Secretary to the King Professor Jean-Louis Vanherweghem Former Chairman of the Board of Directors of the ULB distribution is allowed. distribution

Professor Irina Veretennicoﬀ Professor VUB 9180 only. No further No only. Guest Members Professor Anne De Wit Professor ULB and Scientiﬁc Secretary of the International Committee for Chemistry vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Professor Herv´eHasquin Permanent Secretary of the Royal Academy of Sciences, Letters and Fine Arts of Belgium http://www.sol Professor Franklin Lambert

host on on host Professor VUB Professor Alexander Sevrin Deputy Director for Physics Professor VUB and Scientiﬁc Secretary of the International http://www.worldscientific.com/worldscibooks/10.1142/ Committee for Physics Professor G´ery van Outryve d’Ydewalle Permanent Secretary of the Royal Flemish Academy of Belgium for Sciences and the Arts Professor Lode Wyns Deputy Director for Chemistry ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Director Professor Marc Henneaux

Professor ULB October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg xiii

xiii

Solvay Scientific Committee for Chemistry Professor Kurt Wüthrich (Chair) ETH Zürich, Switzerland and the Scripps Research Institute, La Jolla, USA

Professor Graham Fleming

University of California, Berkeley, USA Professor Robert H. Grubbs, California Institute of Technology, Pasadena, USA Professor Roger Kornberg distribution is allowed. distribution Stanford University, USA

Professor Harold W. Kroto

9180 University of Sussex, Brighton, UK only. No further No only. Professor Henk N.W. Lekkerkerker Utrecht Universiteit, The Netherlands Professor K.C. Nicolaou The Scripps Research Institute, La Jolla, USA vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Professor JoAnne Stubbe Massachusetts Institute of Technology, Cambridge, USA

http://www.sol Professor George M. Whitesides Harvard University, Cambridge, USA

host on on host Professor Ahmed Zewail California Institute of Technology, Pasadena, USA Professor Anne De Wit Universit´e Libre de Bruxelles, Belgium, Scientiﬁc Secretary http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

July 25, 2013 17:28 WSPC - Proceedings Trim Size: 9.75in x 6.5in icmp12-master

This page intentionally left blank distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg xv

Acknowledgements

The organization of the 23rd Solvay Conference has been made possible thanks to the generous support of the Solvay Family, the Solvay Group, the Université distribution is allowed. distribution Libre de Bruxelles, the Vrije Universiteit Brussel, the Belgian National Lottery, the Fédération Wallonie-Bruxelles, the Vlaamse Regering, the Brussels Region, BNP-

9180 Paribas-Fortis, the David and Alice Van Buuren Foundation, Belgacom and the Hˆotel M´etropole. only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg xvi

xvi

Participants Markus Aebi ETH Z¨urich, Switzerland Frances H. Arnold California Institute of Technology, USA

Nenad Ban ETH Z¨urich, Switzerland Jason W. Chin MRC, Cambridge, UK Aled M. Edwards University of Toronto, Canada Joachim Frank Columbia University, USA John A. Gerlt University of Illinois, USA Rudolf Glockshuber ETH Z¨urich, Switzerland distribution is allowed. distribution Adam Godzik Sanford-Burnham Medical Research Institute,

USA Markus G. Grütter University of Zürich, Switzerland 9180 Bernard Henrissat Aix-Marseille University, France only. No further No only. Donald Hilvert ETH Zürich, Switzerland Kendall N. Houk University of California – Los Angeles, USA Judith P. Klinman University of California – Berkeley, USA Henk N.W. Lekkerkerker Utrecht Universiteit, The Netherlands vayinstitutes.be/ Richard M. Lerner The Scripps Research Institute, USA PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Kaspar Locher ETH Z¨urich, Switzerland Tom W. Muir Princeton University, USA

http://www.sol Dario Neri ETH Z¨urich, Switzerland Harry F. Noller University of California – Santa Cruz, USA

host on on host Christine Orengo UCL-IRIS, London, UK Jacques Prost ESPCI, Paris, France Jean-Louis Reymond University of Berne, Switzerland Marina V. Rodnina Max Planck Institute, Germany

http://www.worldscientific.com/worldscibooks/10.1142/ Bryan L. Roth University of North Carolina, USA Sachdev S. Sidhu University of Toronto, Canada Gebhard Schertler Paul Scherrer Institute, Switzerland Peter H. Seeberger Max Planck Institute, Germany Arne Skerra Technische Universit¨at M¨unchen, Germany Raymond C. Stevens The Scripps Research Institute, USA

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Jan Steyaert VIB and Vrije Universiteit Brussel, Belgium Hiroaki Suga The University of Tokyo, Japan NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Craig A. Townsend Johns Hopkins University, USA Wilfred A. van der Donk University of Illinois, USA

Gunnar Von Heijne Stockholm University, Sweden Christopher T. Walsh Harvard Medical School, USA Ian A. Wilson The Scripps Research Institute, USA October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg xvii

xvii

Dennis W. Wolan The Scripps Research Institute, USA Chi-Huey Wong Academia Sinica, Taiwan Kurt W¨uthrich ETH Z¨urich, Switzerland; The Scripps Research Institute, USA

Lode Wyns Vrije Universiteit Brussel, Belgium distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg xviii

xviii

Auditors Antoine Amory Solvay Group Kristin Bartik Universit´e Libre de Bruxelles, Belgium Gilles Bruylants Universit´e Libre de Bruxelles, Belgium

Nico Callewaert Universiteit Gent, Belgium Jean-Fran¸cois Collet Université Libre de Bruxelles, Belgium Stéphanie Deroo Université Libre de Bruxelles, Belgium Anne De Wit Université Libre de Bruxelles, Belgium Jean-Marie Frère UniversitédeLiège, Belgium distribution is allowed. distribution

Erik Goormaghtigh Universit´e Libre de Bruxelles, Belgium

Cédric Govaerts Université Libre de Bruxelles, Belgium Fabrice Homble Université Libre de Bruxelles, Belgium 9180 Denis Lafontaine Université Libre de Bruxelles, Belgium only. No further No only. Anne Lesage GrayMatters Consulting, Belgium Remy Loris Vrije Universiteit Brussel, Belgium Philippe Marion Solvay Group Serge Muyldermans Vrije Universiteit Brussel, Belgium

vayinstitutes.be/ Marc Parmentier Universit´e Libre de Bruxelles, Belgium PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Vincent Raussens Universit´e Libre de Bruxelles, Belgium Han Remaut Vrije Universiteit Brussel, Belgium

http://www.sol Jean-Marie Ruysschaert Universit´e Libre de Bruxelles, Belgium Michel Vandenbranden Universit´e Libre de Bruxelles, Belgium

host on on host Wim Vers´ees Vrije Universiteit Brussel, Belgium http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg xix

xix

Opening Address by Professor Marc Henneaux

Dear colleagues, dear friends, In the name of the International Solvay Institutes, it is my great pleasure to distribution is allowed. distribution welcome all of you to the 23rd Solvay Conference on Chemistry, “New Chemistry and New Opportunities from the Expanding Protein Universe”.

9180 The Conference that starts today is part of a long and prestigious tradition and 2013 is a year that marks the anniversary of many important dates in the history only. No further No only. of the Solvay Institutes. The Institute of Chemistry was created in 1913 by Ernest Solvay. 2013 is therefore the year of its hundredth anniversary. [The Institute of Chemistry fused in 1970 with the Institute of Physics to become The International Solvay Institutes

vayinstitutes.be/ for Physics and Chemistry.] PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- We are celebrating also another anniversary, the 60th anniversary of the 9th Solvay Conference on Chemistry that took place in Brussels in 1953. This conference was quite remarkable because it is during this conference, on the 9th of April, that http://www.sol the original announcement of the double helix structure of DNA deduced by Watson and Crick, was made, prior to the publication of their paper in Nature on April 25, host on on host 1953. The announcement was made by Bragg, who was rapporteur at the conference, and the director of the Cavendish Laboratory in Cambridge where Crick and Watson worked. Bragg had actually very closed ties with the Solvay Institutes since he was http://www.worldscientific.com/worldscibooks/10.1142/ then the chair of the International Committee for Physics, in charge of the Solvay Conferences on Physics. The report by Bragg in the Proceedings contains a note by Watson and Crick, who were clearly aware of the importance of their discovery. We are also celebrating the 30th anniversary of another important conference, the 18th Solvay Conference on Chemistry “Design and Synthesis of Organic Molecules Based on Molecular Recognition”. In that conference, topics related to this year’s conference were discussed. And indeed at least two of this year’s partic- ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World ipants were already participants in 1983. This was an excellent conference, and a

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY signiﬁcant fraction of the participants got the Nobel prize in the meantime! If I have chosen to mention these two conferences, that of 1953 and that of 1983, it is, besides the fact that they both occurred a multiple of ten years ago, because

proteins played a central role in them. Proteins are again the stars of this year’s conference. This is not a topic that I know well — I am a physicist — but while reading the Proceedings of the previous conferences on the subject and preparing October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg xx

xx Opening Address

my speech, I have been impressed and amazed by the beauty and intricacy of the discipline. I am convinced that the research carried in the last sixty years has not only brought an enormous, overwhelming understanding of the growing protein

universe, which might be exponentially expanding if I understand well, but also has

led to new questions, new puzzles and new challenges which are as fascinating and intriguing as the challenges that existed in the ﬁfties. As you know, the Solvay Conferences are very special. These are elitist conferences by invitation-only, with a limited number of participants. There are few presentations but a lot of discussions. People come to the Solvay Conferences for the scientiﬁc interactions, which are indeed privileged, not for giving a talk. The distribution is allowed. distribution format adopted this year to implement this philosophy and to initiate the discus-

sions is an innovation with respect to the original format where each session started with one or two rapporteur talks. The rapporteur talks have been replaced by 5 to 9180 6 short introductory statements. only. No further No only. For the discussions to be fruitful, a careful preparation is needed. Here is how it goes: the subject and scientiﬁc organizers of the Solvay Conferences are chosen by the Solvay International Scientiﬁc Committee for Chemistry, which has complete freedom in doing so. The organizers are then in charge of the invitations and the

vayinstitutes.be/ program, and again have complete “carte blanche” for achieving this task. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- I would like to express our deepest thanks to the Solvay Scientific Committee for Chemistry, and in particular to its chair, Kurt Wüthrich, for choosing the subject of the 23rd Solvay Conference on Chemistry. I would also like to express our gratitude http://www.sol to Ian Wilson, Donald Hilvert and Kurt Wüthrich, for accepting to organize the Conference. I know that this means an enormous amount of work. host on on host Since the discussions are important, they are included in the Proceedings. This is the tradition. We have an editorial committee that already remarkably worked since all the contributions have been received before the meeting. This is to be praised and I would like to thank the editorial committee, and in particular Anne

http://www.worldscientific.com/worldscibooks/10.1142/ De Wit and Dennis Wolan, for carrying the publication task. They are assisted for the transcriptions of the discussions by a scientiﬁc secretariat to whom we are also very grateful. Thank you very much for your attention. I wish you a very fruitful meeting.

Marc Henneaux, Director of the Solvay Institutes ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg xxi

xxi

Preface

On the occasion of the 150th anniversary of the Solvay Company and the centen- nial anniversary of the ﬁrst Chemistry Council in 1913, the November/December distribution is allowed. distribution 2013 issue of Chemistry International devoted its cover to “A Look Back at Ernest Solvay”, and pages 4–11 to this same theme and to “The Solvay Chemistry Coun-

9180 cils”. It is thus highly ﬁtting that a Solvay Conference on Chemistry was held in the same year, with the theme “New Chemistry and New Opportunities from the only. No further No only. Expanding Protein Universe”. This 23rd Solvay Conference on Chemistry assembled an illustrious group of invited scientists representing a diverse range of research disciplines, including chemistry, chemical biology, computational biology, structural biology, biochemistry, and cell biology. All of the attendees are at the forefront vayinstitutes.be/ of their ﬁelds and have advanced our knowledge of the structure and function of PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- biological macromolecules and their supramolecular assemblies, the chemistry of essential biological processes, and the development of new small-molecule drugs and biologicals. http://www.sol The theme of this Solvay Conference was inspired by explorations into the expanding protein universe, using the tools of proteomics and structural genomics host on on host that have opened up new frontiers and opportunities for scientiﬁc investigation. The program was presented in six Sessions, and the Proceedings are organized accordingly. Session 1 on “New Chemistry in the Expanding Protein Universe” was devoted to innovative reports on discovering new chemistry associated with explo- http://www.worldscientific.com/worldscibooks/10.1142/ ration of today’s expanded protein space. Session 2 on “Exploring Enzyme Families and Enzyme Catalysis” was focused on recently identiﬁed families of enzymes and on studies of enzyme mechanisms. Session 3 on “Microbiomes and Carbohydrate Chemistry” was devoted on the one hand to the microbiota that are adapted to their particular environment in the human gut, and consequently have enhanced networks for carbohydrate metabolism, and on the other hand to recent advances in carbohydrate synthetic chemistry. Session 4 on “GPCRs and Transporters: ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Ligands, Cofactors, Drug Development” was dominated by exciting new results

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY on GPCRs and membrane transporters from structural biology and biophysics, with a strong emphasis on biomedical applications. Session 5 on “Biologicals and Biosimilars” provided a perspective on the current state of development of protein-

based drugs. The ﬁnal Session 6 on “Proteins in Supramolecular Machines” focused on the impressive recent developments in understanding and dissecting complex macromolecular assemblies with the use of structural biology. The six Sessions thus October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg xxii

xxii Preface

provided a spectacular overview of the present state of protein science and an outlook to anticipated developments during the coming years and in the more distant future.

Considering the wide scope of present-day protein science, we decided on a

novel format for this Conference. Instead of having one “Rapporteur” providing a one-hour introduction to each Session, we scheduled short “Personal Statements” by the Session Chairperson and six to eight Panel Members at the outset of each Session. These presentations segued into a discussion among the designated Panel Members, which initiated lively exchanges with the Plenum. The Proceedings are organized in such a way that for each Session the short “Personal Statements”, distribution is allowed. distribution which had been submitted prior to the start of the Conference, are followed by

the transcription of the ensuing discussion. This Conference format was designed to promote thought-provoking questions and answers that are not usually found 9180 in formal presentations. The Conference participants were great in accepting this only. No further No only. rather unusual format. As a result, the combination of the “Personal Statements” and the discussion transcriptions for each Session now present unique snapshots of present-day protein science and outlooks to the future. Protein science presented an ideal foundation for the organization of the “First

vayinstitutes.be/ Solvay Roundtable on the State of Science & Present Challenges: Biologicals and PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Biosimilars” on one of the afternoons of the Conference week. The Roundtable assembled a group of industry leaders with Conference participants taken primarily from among the speakers in Session 5. The following format was proposed by Jean- http://www.sol Marie Solvay: “The aim of the roundtable is to enable an informal discussion of the state-of-the-art on biologicals and biosimilars, making it possible for outstanding host on on host scientists participating in the 23rd Solvay Conference on Chemistry to share their views with interested business leaders”. The envisaged aim was achieved due to the mutual interest in the theme for the Roundtable, giving rise to intense discussions during the afternoon and thereafter over dinner.

http://www.worldscientific.com/worldscibooks/10.1142/ The culmination of the Conference week was the Public Event held by the Inter- national Solvay Institutes at the Flagey Cultural Center. The well-attended event began with the Solvay Award Ceremony for young scientists working at local uni- versities in the ﬁelds of physics, chemistry and engineering. This was followed by lectures of Professors Joachim Frank (Columbia University, USA) and Jason Chin (University of Cambridge, UK) on their current research. The afternoon concluded with a debate prompted by a wide range of excellent questions posed by the audi-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World ence to a panel of Conference attendees. It was a pleasure for me to chair this panel

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY of Professors Jason Chin (MRC, Cambridge, UK), Joachim Frank (Columbia Uni- versity, USA), Gunnar von Heijne (Stockholm University, Sweden), Donald Hilvert (ETH Z¨urich, Switzerland), Christopher Walsh (Harvard Medical School, USA) and Ian A. Wilson (The Scripps Research Institute, USA). I would like to thank Prof. Franklin Lambert for his role as bilingual moderator, and the Solvay Institutes for hosting this well-received and successful event of public outreach. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg xxiii

Preface xxiii

It was a privilege for me to chair this Solvay Conference on Chemistry, which gave me and my colleagues, Prof. Donald Hilvert and Prof. Ian A. Wilson, the opportunity to select a stellar group of scientists from diﬀerent disciplines for a

truly integrative discussion on timely subjects. We greatly appreciate the generous

support provided by the Solvay Institutes. I was impressed by the commitment of the Solvay family, with Jean-Marie Solvay participating in the scientiﬁc part of the Conference and being joined by Mrs. Solvay at the social functions. We want to thank the Solvay Institutes represented by Prof. Marc Henneaux. Special thanks go to Dominique Bogaerts and Isabelle Van Geet from the Solvay Institutes, for their commitment and dedicated assistance that made the logistical running of distribution is allowed. distribution the conference perfect, to Prof. Anne De Wit, who arranged the recording and

transcription of the discussions during the meeting, and to the local scientists who aided with the transcriptions. 9180 only. No further No only. Kurt W¨uthrich Chair of the Conference vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

July 25, 2013 17:28 WSPC - Proceedings Trim Size: 9.75in x 6.5in icmp12-master

This page intentionally left blank distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 1

Session 1 New Chemistry in the Expanding Protein Universe distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 2

distribution is allowed. distribution

Biosynthesis of nisin A, a lantibiotic that has been used for nearly

9180 50 years in the food industry to combat food-borne pathogens. See Figure 2 contributed by Wilfred van der Donk on page 8. only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 3

NOVEL CHEMISTRY STILL TO BE FOUND IN NATURE

CHRISTOPHER T. WALSH Biochemistry & Molecular Pharmacology Department Harvard Medical School, Boston, MA 02115, USA distribution is allowed. distribution

My view of the present state of research on new chemistry in the expanding protein universe 9180 New chemical transformations catalyzed by proteins continue to be discovered. only. No further No only. Most of the novel scaﬀolds in small molecule frameworks emerge from studies on microbes, prokaryotes and single cell eukaryotes, from underexplored niches with both anaerobic and aerobic (oxidative) metabolic transformations prominent in bond-breaking and bond-making steps. Bioinformatic analysis of protein su-

vayinstitutes.be/ perfamilies may highlight particular family members for novel catalytic activities. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Post-translational modiﬁcation of ribosomally generated proteins has also been implicated in the morphing of peptide frameworks into complex architectures. De novo design and protein evolution activity can also create novel chemical transformations, http://www.sol including reactions not previously seen in Nature. host on on host My recent contributions to new chemistry in the expanding protein universe Research from my group has focused on the morphing and maturation of peptide

http://www.worldscientific.com/worldscibooks/10.1142/ scaffolds into rigidified, compact scaffolds by the two major biosynthetic strategies for peptide bond formation: ribosomal and non-ribosomal assembly lines. A hallmark of the RNA-independent nonribosomal peptide synthetases has been the use of nonproteinogenic amino acid building blocks in place of the 20 canonical proteinogenic amino acids. As an example, we have examined how a subset of mononuclear nonheme iron oxygenases act instead as halogenating catalysts to provide γ-chloro amino acid and cyclopropyl amino acid building blocks to NRPS assembly lines.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World The nonproteinogenic β-amino acid anthranilate is a building block for a series of fungal alkaloids ranging from bicyclic to octacyclic scaﬀolds, put together by bi- and NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY trimodular NRPS assembly lines, followed by action of dedicated tailoring enzymes. Complexity generation in peptide scaﬀolds can also be achieved from ribosomally

generated nascent proteins by a series of post-translational modiﬁcations (PTMs). A remarkable cascade of more than a dozen PTMs occur as a 14 residue C-terminal peptide region of a 52mer is morphed into the trithiazolylpyridine core of thia- zolyl peptide antibiotics in the thiocillin and nosiheptide class of antibiotics. These October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 4

4 C. T. Walsh

involve conversion of cysteines to thiazoles, threonines to methyloxazoles and partitioning of serine residues down two distinct PTM pathways. One route is to the corresponding oxazole by cyclization, dehydration, and aromatization; the other

route is net dehydration to dehydroalanines. A pair of dehydroalanines can un-

dergo condensation and dehydrative aromatization to yield the core pyridine ring of the 2,4,6-trithiazolyl pyridine core at the center of these mature antibiotic scaﬀolds. distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

http://www.sol Fig. 1. Complexity generation in peptide scaﬀolds: two strategies. host on on host Outlook to future developments of research in the chemistry of the expanding protein universe As bioinformatics, structural genomics and proteomics continue to deﬁne the ex-

http://www.worldscientific.com/worldscibooks/10.1142/ isting protein universe and guide protein engineers in search of new kinds of catalysts, the future is bright for novel chemistry to continue to emerge. On one end of the catalytic spectrum the large superfamily of radical S-adenosyl methio- nine (SAM) enzymes holds particular promise for catalytic diversity and novelty of chemical transformations: the reactions catalyzed by such famly members as riibonucleotide reductase, lipoyl and biotin synthases, ThiiC, and the tRNA and mRNA C-methyltransferases are likely to be the tip of the iceberg in the chemical

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World capacity of this superfamily. In a distinct superfamily, the hemeprotein cyochrome

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY P450 oxygenases, protein engineering has recently led to evolution of synthetically useful carbene chemistry. De novo protein design should enable completely abiotic chemistry to move into the realm of protein catalysis.

References 1. C. T. Walsh, R. O’Brien, C. Khosla, Angew. Chem. Int. Ed. 52, 7098 (2013). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 5

Novel Chemistry Still to be Found in Nature 5

2. C. T. Walsh, S. Haynes, B. Ames, X. Gao, Y. Tang, ACS Chem. Biol. 8, 1366 (2013). 3. C. T. Walsh, S. Malcolmson, T. Young, ACS Chem. Biol. 7, 429 (2012).

4.P.S.Coelho,J.Wang,M.E.Ener,S.A.Baril,A.Kannanet al., Nat. Chem.

Biol. 9, 485 (2013). distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 6

NATURAL PRODUCT BIOSYNTHESIS IN THE GENOMIC AGE

WILFRED A. VAN DER DONK Department of Chemistry, University of Illinois at Urbana-Champaign and the Howard Hughes Medical Institute, 600 South Mathews Ave, Urbana, IL 61801, USA distribution is allowed. distribution

My view of the present state of research on new chemistry in the expanding protein universe 9180 Natural products (NPs) have featured prominently in the development of pharma- only. No further No only. ceuticals and as tools to study biology. However, since the turn of the century, many natural product discovery platforms in industry have been dissembled. Sev- eral explanations have been given for the withdrawal of pharmaceutical companies from NP discovery. For antibiotics, small projected proﬁts have been a major

vayinstitutes.be/ driving force. But when considering e.g., antitumor agents, the move away from PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- NPs has been motivated by other factors including high rediscovery rates of known compounds and the diﬃculty to perform medicinal chemistry because of complex structures. At the same time, the available sequenced genomes have demonstrated http://www.sol that the number of NP biosynthetic gene clusters in a typical microorganism far exceeds the number of compounds it produces under laboratory conditions. Based host on on host on their sequences, the overwhelming majority of these “silent” gene clusters are expected to encode new NPs. As a result, genome mining has been widely touted as a potentially eﬃcient route to new NPs. An important bonus of investigating NP biosynthetic pathways with respect to the theme of this panel is their richness

http://www.worldscientific.com/worldscibooks/10.1142/ in novel biochemical transformations.

My recent research contributions to new chemistry in the expanding protein universe The genome sequencing eﬀorts have revealed that ribosomally synthesized and post- translationally modiﬁed peptides (RiPPs) form a much larger class of NPs than

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World anticipated [1]. The extensive post-translational modifications endow these peptides with greatly expanded chemical structures, restricted conformational flexibility for NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY target recognition, and increased metabolic stability. In retrospect, it may not be surprising that RiPP biosynthetic pathways are so widespread because they offer a

unique advantage to their producing organisms: high evolvability. Nearly all RiPPs are initially synthesized as a longer precursor peptide, with a leader peptide appended to the N-terminus of the core peptide, which will be converted to the ﬁnal NP (Fig. 1(a)). The leader peptide is important for recognition October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 7

Natural Product Biosynthesis in the Genomic Age 7

by many of the post-translational modiﬁcation enzymes [2]. This leader peptide- guided strategy results in highly evolvable pathways because the post-translational processing enzymes can be intrinsically permissive with respect to mutations in the

core peptide. The leader peptide-guided biosynthesis also renders RiPPs particu-

larly well-suited to both genome mining and synthetic biology approaches because their substrates are DNA encoded and the biosynthetic enzymes have intrinsically relaxed substrate speciﬁcity [3]. distribution is allowed. distribution

9180 only. No further No only.

(a) (b)

vayinstitutes.be/ Fig. 1. Biosynthesis of RiPPs. (a) General biosynthetic pathway of RiPPs. (b) General biosyn- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

thetic strategy that results in the formation of thioether corsslinks named lanthionine and methyl- - lanthionine.

http://www.sol Based on the available genomes, lanthionine-containing peptides (lanthipeptides) are the most abundant class of RiPPs. Currently known lanthipeptides

host on on host have a range of activities including antimicrobial (called lantibiotics), morpho- genetic, antiviral, and antiallodynic [4]. Lanthionine (Lan) consists of two alanine residues crosslinked via a thioether linkage that connects their β-carbons; 3- methyllanthionine (MeLan) contains one additional methyl group (Fig. 1(b)). These

http://www.worldscientific.com/worldscibooks/10.1142/ structures are installed by a dehydratase that eliminates water from Ser and Thr residues to generate dehydroalanine (Dha) and dehydrobutyrine (Dhb), respectively, and a cyclase that catalyzes the subsequent addition of thiols of Cys residues to the dehydro amino acids to generate Lan and MeLan, respectively. How the enzymes coordinate these complex chemical transformations in which the substrate peptide structure is continuously changing is unknown (e.g., Fig. 2). Furthermore, most lanthipeptide biosynthetic enzymes do not have homology with non-lanthipeptide

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World proteins, and hence many questions remain about their evolutionary origin. At least four different pathways to these polycyclic natural products have evolved NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY [4], reflecting the high efficiency and evolvability of a post-translational modification route to generate conformationally constrained peptides. Recent studies have

shown that three of the four pathways involve Ser/Thr phosphorylation and sub- sequence phosphate elimination to generate the dehydro amino acids [6–8]. The fourth route unexpectedly involves Ser/Thr glutamylation [9]. As can be seen in Fig. 2, the dehydratase and cyclase enzymes involved in lanthipeptide biosynthesis October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 8

8 W. A. van der Donk

distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Fig. 2. Biosynthesis of nisin A, a lantibiotic that has been used for nearly 50 years in the food industry to combat food-borne pathogens. Note the challenge for the cyclase to regioselectively connect ﬁve Cys nucleophiles with the correct dehydro amino acids, a process that can generate

http://www.sol more than 6,500 different constitutional isomers [5]. In this process, five rings that differ greatly in size and amino acid sequence are formed by one enzyme. host on on host actonresiduesthatarelocatedindiversesequence contexts. These lanthionine synthetases appear to have retained the low substrate specificity of their primitive progenitors but acquired a dependence on a leader peptide. The exact role of the

http://www.worldscientific.com/worldscibooks/10.1142/ leader peptide has been widely debated, and no definitive answer has been provided thus far. Our studies suggest that the leader peptide plays an allosteric role [5], but other roles cannot be ruled out and it is possible that the leader peptides have different functions in different classes of RiPPs. A recent genome mining exercise revealed a stunning example of natural combinatorial biosynthesis, illustrating the considerable potential of RiPP biosynthesis for synthetic biology. The genome of a strain of Prochlorococcus, a planktonic ma-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World rine photosynthetic cyanobacterium, encodes a single class II lanthionine synthetase (ProcM) but no less than 29 diﬀerent putative substrate peptides. These substrates NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY have highly conserved leader peptides but hypervariable core peptides (not a single homologous pair). We cloned the enzyme and a subset of its putative substrates

and showed that all 17 peptides tested were indeed substrates for ProcM [10]. In a follow-up study, we determined the structure of a subset of the resulting compounds termed prochlorosins (Pcn), and demonstrated that their ring topology is highly diverse [11]. These ﬁndings open up a large number of intriguing questions, October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 9

Natural Product Biosynthesis in the Genomic Age 9

including (1) how can one enzyme make 29 very diﬀerent polycyclic structures, (2) what is the function of these cyclic peptides, and (3) can the ProcM-activity be harnessed for synthetic purposes?

Outlook to future developments of research on new chemistry in the expanding protein universe Genome mining eﬀorts are highly likely to uncover many new NPs. However, genome mining approaches have their own challenges. Most importantly, unlike phenotypic screens that provide compounds with a desired activity, tools to predict

distribution is allowed. distribution the type of bioactivity of a putative new compound identiﬁed by genome mining

are lacking. Hence, for genome mining to be a viable approach to new compounds

with desirable activities, new predictive tools need to be developed and/or high- 9180 throughput production platforms must be engineered such that the odds of ﬁnding molecules with desired activities are increased. Eﬃcient production methods could only. No further No only.

either be based on production in heterologous hosts or on new strategies to elicit production from the original organisms [12]. RiPPs lend themselves particularly well for heterologous expression as the 20 amino acids are common building blocks that are present in any heterologous hosts and because their substrates are gene vayinstitutes.be/ encoded. In addition, synthetic biology approaches may take advantage of the high PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- substrate tolerance of RiPP biosynthetic enzymes to generate cyclic peptide libraries that can be selected for speciﬁc properties. http://www.sol Acknowledgments host on on host This research was supported by the National Institutes of Health (RO1 GM 58822) and the Howard Hughes Medical Institute.

References http://www.worldscientific.com/worldscibooks/10.1142/ 1. P. G. Arnison, M. J. Bibb, G. Bierbaum, A. A. Bowers, T. S. Bugni et al., Nat. Prod. Rep. 30, 108 (2013). 2. T. J. Oman, W. A. van der Donk, Nat. Chem. Biol. 6, 9 (2010). 3. J. E. Vel´asquez, W. A. van der Donk, Curr. Opin. Chem. Biol. 15, 11 (2011). 4. P. J. Knerr, W. A. van der Donk, Annu. Rev. Biochem. 81, 479 (2012). 5. X. Yang, W. A. van der Donk, Chem. Eur. J. 19, 7662 (2013).

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 6. C. Chatterjee, L. M. Miller, Y. L. Leung, L. Xie, M. Yi et al., J. Am. Chem. Soc. 127, 15332 (2005). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 7. W. M. M¨uller, T. Schmiederer, P. Ensle, R. D. S¨ussmuth, Angew. Chem. Int. Ed. 49, 2436 (2010).

8.Y.Goto,B.Li,J.Claesen,Y.Shi,M.J.Bibbet al., PLoS Biol. 8, e1000339 (2010). 9. N. Garg, L. M. Salazar-Ocampo, W. A. van der Donk, Proc. Natl. Acad. Sci. USA 110, 7258 (2013). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 10

10 W. A. van der Donk

10. B. Li, D. Sher, L. Kelly, Y. Shi, K. Huang et al., Proc. Natl. Acad. Sci. USA 107, 10430 (2010). 11. W. Tang, W. A. van der Donk, Biochemistry 51, 4271 (2012).

12. K. Scherlach, C. Hertweck, Org. Biomol. Chem. 7, 1753 (2009).

distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 11

PEPTIDE DENDRIMERS AND POLYCYCLIC PEPTIDES

JEAN-LOUIS REYMOND Department of Chemistry and Biochemistry, University of Bern, Freiestrasse 3 3012-Bern, Switzerland distribution is allowed. distribution

My view of the present state of research on new chemistry in the expanding protein universe 9180 Proteins are classically understood as the products of genes as found in Nature. only. No further No only. Advances in protein engineering such as mutagenesis, directed evolution, and non- natural amino acid mutagenesis have allowed to not only better understand how proteins work, but also to produce new proteins with new functions. The bulk of these advances were made possible by the techniques of molecular biology. Synthetic

vayinstitutes.be/ chemistry follows closely behind in exploring the protein universe using solid-phase PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- peptide synthesis (SPPS) [1] and native chemical ligation (NCL) [2], and contributes by allowing the chemical synthesis of proteins as well as the introduction of targeted modiﬁcations that cannot be produced in a cell. http://www.sol

host on on host My recent research contributions to new chemistry in the expanding protein universe Our approach to de novo protein design focuses on rethinking peptide topologies from ﬁrst principles, intentionally departing from Nature’s models in search for

http://www.worldscientific.com/worldscibooks/10.1142/ new protein-like macromolecules with unexpected properties. In the hierarchical organization of organic matter, atoms assemble in topologically diverse patterns of covalent bonds (graphs) to form millions of diﬀerent molecules [3]. At the next level amino acids assemble through peptide bonds to form peptides and proteins, however in this case topological diversity is limited almost exclusively to linear peptide chains as encoded in genes and translated by the ribosome. Proteins acquire their diverse shapes and functions not through topologically diverse peptide bond networks but

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World rather by folding. One advantage of chemical synthesis is that one is not limited to linear chains when assembling a polypeptide, and one can also access diﬀerent NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY topologies by introducing multiple branching points in form of diamino acids or amino diacids. Capitalizing on this additional degree of freedom, we have developed

SPPS protocols to prepare artificial proteins with non-natural topologies. In one such approach inspired by the field of organic dendrimers [4], we are investigating peptide dendrimers [5] as artificial proteins [6]. Peptide dendrimers are symmetrically multi-branched peptides not accessible by biosynthetic routes. We October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 12

12 J.-L. Reymond

(a) (b) a) HO OH HO OH b) O O HO OH HO O HO O O OH OH HO O OH HO OH

O O O HN HN HO O O Lys OH HN Lys

Pro Lys Pro Leu Pro Leu O Lys HN Leu Phe Pro Lys Lys Lys Leu Ile Phe Ile Lys IleNH2 Lys His

(c)c) distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

http://www.sol Fig. 1. The glycopeptide dendrimer P. aeruginosa bioﬁlm inhibitor GalAG2 is shown in (a) chemical structure, (b) MD minimized model and (c) modelled complex with its target LecA. host on on host install branching points every 2nd,3rd or 4th amino acid along the growing peptide chain in form of diamino-acids (lysine or 2,3-diaminopropanoic acid) resulting in G1 (one branching point, two terminal branches), G2 (three branching points, four

http://www.worldscientific.com/worldscibooks/10.1142/ terminal branches), or G3 (seven branching points, eight terminal branches) peptide dendrimers with variable amino acid positions throughout the dendrimer branches. This divergent dendrimer SPPS delivers products in excellent yields and purities for sequences of up to approximately fourty residues in total, corresponding to twelve consecutive coupling steps. The method is amenable to combinatorial split-and- mix synthesis to produce libraries of thousands of diﬀerent dendrimers which can be screened for function by on-bead or oﬀ-bead assays [7, 8]. We have discov-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World ered peptide dendrimers with a variety of functions including enzyme-like catalysis (esterase [9], aldolase [10], cobalamin binding proteins [11], peroxydase [12]), an- NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY timicrobial activity [8], and bioﬁlm inhibition using glycopeptide dendrimers lectin inhibitors (Fig. 1) [13]. We also found that certain polycationic/lipophilic peptide

dendrimers acts as highly eﬃcient yet non-toxic gene transfer reagents [14], while others can deliver small molecule cargos such as cytotoxic agents into cells [15, 16]. One remarkable feature of peptide dendrimers is their increased proteolytic stability compared to linear peptides, probably caused by steric crowding around the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 13

Peptide Dendrimers and Polycyclic Peptides 13

peptide bonds [17]. Contrary to linear peptides which are frequently prone to aggregation, peptide dendrimers are generally well behaved in solution and do not aggregate. Structural investigations by CD, FTIR and NMR spectroscopy and by

molecular dynamics simulations show that peptide dendrimers are conformation-

ally ﬂexible and adopt a globular molten globule like structure with relatively few peptide backbone H-bonds and consequently few secondary structure elements, although α-helical peptide dendrimers can be obtained by design [18, 19]. The direct dendrimer SPPS yields pure products up to G3 dendrimers and ca. 5 kDal in size. We have explored chemical ligation methods to prepare higher generation, protein sized dendrimers through convergent assembly of G2 and G3 den- distribution is allowed. distribution drimers. While NCL and click reactions were unsuccessful in our hands, protein-

sized dendrimers were obtained by the coupling of multiple copies of a dendrimer carrying a cysteine residue at its core with another dendrimer substituted with four 9180 or eight chloroacetyl groups at its N-termini. This so-called ClAc (Chloroacetyl only. No further No only. thioether) ligation allowed us to prepare peptide dendrimers up to 340 amino acids characterized as enzyme models [10]. In a comparable approach to artiﬁcial proteins, we are investigating the synthesis and properties of topologically diverse polycyclic peptides designed as peptide

vayinstitutes.be/ analogs of saturated alkane graphs [20]. We have established an SPPS protocol PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- comprising on-bead cyclisation, purification of the cyclic peptide, and closing of the second ring in solution and used the approach to prepare bicyclic peptide analogs of “non-zero bridged” bicyclic alkanes such as norbornane (Fig. 2). http://www.sol Bicyclic peptides define a very large structural family. For instance, there are 25 possible bicyclic peptide scaffolds up to 10 amino acids forming 3.3.1012 possible host on on host products using 20 different amino acids at the variable non-bridgehead positions. http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Fig. 2. Experimental MR structures of selected bicyclic peptides. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 14

14 J.-L. Reymond

Analysis of molecular shape shows that bicyclic peptides populate a distinct area of peptide shape space. Thus, while linear peptides in extended or β-sheet conformation are rod-like and monocyclic peptides are disk-like, “non-zero bridged” bicyclic

peptides have a signiﬁcant third molecular dimension indicative of a globular shape,

and “zero-bridged” bicyclic peptides are elliptical. The shapes of bicyclic peptides resemble those of short α-helical segments extracted from proteins in the PDB. Bi- cyclic peptides are conformationally rigid and thus overcome the major challenge encounterd sofar in our peptide dendrimer work. Bicyclic peptide have the potential for selective protein binding, and are one of the current focus of our research. distribution is allowed. distribution

Outlook to future developments of research on new chemistry in

the expanding protein universe 9180 The chemical synthesis of peptides by SPPS is suﬃciently versatile and practical only. No further No only.

to allow the rapid exploration of functional sequence space, in particular through the use of parallel or combinatorial synthesis. We are exploiting this technique to explore non-natural topologies deﬁning entirely new classes of macromolecules, yet with the same building block composition as proteins and therefore a particular

vayinstitutes.be/ propensity for biocompatibility. In the near future we are particularly interested to PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- integrate computational modelling for the design of peptide dendrimers and polycyclic peptides, and thus replace untrained combinatorial screening with educated guesses and the synthesis of focused series. If modelling delivers on its promise this

http://www.sol approach should allow a more focused exploration of protein sequence and topological space outside of what Nature has realized sofar. host on on host

Acknowledgments This work was supported ﬁnancially by the University of Berne, the Swiss National

http://www.worldscientific.com/worldscibooks/10.1142/ Science Foundation, and the EU ITN BioChemLig.

References 1. B. Merriﬁeld, Meth. Enzymol. 289, 3 (1997). 2. S. B. Kent, Chem. Soc. Rev. 38, 338 (2009). 3. L. Ruddigkeit, R. Van Deursen, L. C. Blum, J. L. Reymond, J. Chem. Inf.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Model. 52, 2864 (2012). 4. D. Astruc, E. Boisselier, C. Ornelas, Chem. Rev. 110, 1857 (2010). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 5. L. Crespo, G. Sanclimens, M. Pons, E. Giralt, M. Royo et al., Chem. Rev. 105, 1663 (2005).

6. J.-L. Reymond, T. Darbre, Org. Biomol. Chem. 10, 1483 (2012). 7. N. Maillard, A. Clouet, T. Darbre, J. L. Reymond, Nat. Protoc. 4, 132 (2009). 8. M. Stach, N. Maillard, R. U. Kadam, D. Kalbermatter, M. Meury et al., Med- ChemComm. 3, 86 (2012). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 15

Peptide Dendrimers and Polycyclic Peptides 15

9. R. Biswas, N. Maillard, J. Kofoed, J. L. Reymond, Chem. Commun. 46, 8746 (2010). 10. N. A. Uhlich, T. Darbre, J.-L. Reymond, Org. Biomol. Chem. 9, 7071 (2011).

11. N. A. Uhlich, A. Natalello, R. U. Kadam, S. M. Doglia, J. L. Reymond

et al., ChemBioChem. 11, 358 (2010). 12. P. Geotti-Bianchini, T. Darbre, J. L. Reymond, Org. Biomol. Chem. 11, 344 (2013). 13. J. L. Reymond, M. Bergmann, T. Darbre, Chem. Soc. Rev. 42, 4814 (2013). 14. A. Kwok, G. A. Eggimann, J. L. Reymond, T. Darbre, F. Hollfelder, ACS Nano 7, 4668 (2013). distribution is allowed. distribution 15. E. M. Johansson, J. Dubois, T. Darbre, J. L. Reymond, Bioorg. Med. Chem.

18, 6589 (2010). 16. G. A. Eggimann, S. Buschor, T. Darbre, J. L. Reymond, Org. Biomol. Chem. 9180 11, 6717 (2013). only. No further No only. 17. P. Sommer, V. S. Fluxa, T. Darbre, J. L. Reymond, ChemBioChem. 10, 1527 (2009). 18.S.Javor,A.Natalello,S.M.Doglia,J.L.Reymond,J. Amer. Chem. Soc. 130, 17248 (2008).

vayinstitutes.be/ 19. H. K. Ravi, M. Stach, T. A. Soares, T. Darbre, J. L. Reymond et al., Chem. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Commun. 49, 8821 (2013). 20. M. Bartoloni, R. U. Kadam, J. Schwartz, J. Furrer, T. Darbre et al., Chem. Commun. 47, 12634 (2011). http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 16

WHAT CAN COMPARATIVE GENOMICS REVEAL ABOUT THE MECHANISMS OF PROTEIN FUNCTION EVOLUTION?

NATALIE L. DAWSON∗,ROMAINSTUDER∗, NICK FURNHAM†, DAVID LEES∗, SAYONI DAS∗, JANET THORNTON† and CHRISTINE ORENGO∗ ∗University College London, London, UK †European Bioinformatics Institute, Hinxton, UK distribution is allowed. distribution

9180 My view of new chemistry in the expanding protein universe

only. No further No only. Over the last decade, international genomics initiatives have a brought a ﬂood of

new protein sequence and structural data, increasing our knowledge of the protein family and functional repertoire across the tree of life, particularly for microbial organisms. In parallel with major developments in sequencing technologies, the worldwide structural genomics initiatives have brought improved strategies for solv- vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS ing the 3D structures of proteins and have expanded our knowledge of the universe

- of protein folds [1]. Whilst this data is still much more sparse than the sequence data, methods for mapping sequences to known structures have advanced and have

http://www.sol facilitated the comparison of functional sites between related proteins. Currently, there are over a thousand evolutionary protein enzyme families characterized both

host on on host structurally and by sequence, of which ∼ 100 have been very widely exploited during evolution [2]. Since the 3D locations of the active sites are frequently conserved across a family, by comparing the diﬀerences in physico-chemical characteristics of residues in the active sites of relatives it is possible to recognize those which have evolved novel chemistries. Metagenomic studies are revealing great diversity http://www.worldscientific.com/worldscibooks/10.1142/ in the chemical repertoires of bacteria which should help in understanding enzyme mechanisms and designing novel catalysts.

Recent research contributions to new protein chemistry Some domain superfamilies are very ancient and highly populated [3]. Duplication and divergence in these superfamilies has played a major role in expanding the ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World functional repertoires of species. My group has been analysing, computationally,

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY the ways in which divergence in sequence and structure can modify protein function.

Identifying functionally distinct relatives within a protein domain

superfamily Identifying functionally distinct homologues from the sequence data is a challenging computational problem as extensive residue insertions between very diverse October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 17

What can Comparative Genomics Reveal About Protein Evolution? 17

relatives means that it is often hard to align all sequences in the robust manner needed to derive a phylogenetic tree. FunFHMMer which classifies functional families in our CATH classification [4, 5] uses sensitive profile-profile based methods to

detect conserved sequence patterns highly likely to be shared by functionally similar > homologues. Classiﬁation of 2,500 completed genomes and UniProt ( 25 million sequences in total), gives ∼ 50,000 diﬀerent functional families, 30,000 of which have experimentally characterised functions.

Using structural data to derive phylogenetic trees that reveal the evolution of new functions in superfamilies distribution is allowed. distribution

Because the 3D structures of proteins are much more conserved than their sequences,

structural data can guide the alignment of relatives to generate phylogenetic trees. 9180 Together with Janet Thornton at the EBI, we are using these trees to study the emergence of new functions in enzyme superamilies. The FunTree resource devel- only. No further No only.

oped by Nick Furnham [6] integrates phylogenetic information with a host of data collected by the Thornton group on the substrates, chemistry and reaction mechanisms performed by each relative, together with information on the catalytic and binding residues involved. This treasure trove of functional data can help reveal the vayinstitutes.be/ mechanisms leading to functional changes; insights that can be valuable for protein PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- design.

http://www.sol To what extent can function diverge within a protein superfamily Using FunTree and stepping down through the evolutionary tree in each superfamily, host on on host from the root to the leaves, it is clear that dramatic changes in the chemical class of the reaction performed are rare, although possible [6].

Changes in active site residues can modify the chemistry http://www.worldscientific.com/worldscibooks/10.1142/ performed Dramatic changes in chemistry are likely to be mediated by variations in the physiochemical characteristics of residues lining the active site pockets, which are either directly involved in the reaction mechanism or contribute to an electrostatic environment that promotes certain types of reactions. We used structural alignments to identify equivalent positions in the active sites of diverse functional relatives within

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World a superfamily. Figure 1 shows the extent to which relatives can diverge in physicochemical properties of residues at these sites and also shows the number of diﬀerent NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY enzyme chemistries identiﬁed for relatives in these superfamilies. In some superfamilies, there are clearly positions in the structure where mu-

tations more frequently lead to functional change, whilst other superfamilies show plasticity across the whole active site pocket [7]. A large proportion of enzyme superfamilies adopt TIM and Rossmann folds, which have structural frameworks that support huge sequence diversity. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 18

18 N. L. Dawson et al.

distribution is allowed. distribution

9180 only. No further No only.

Fig. 1. Plot showing the average similarity in physicochemical properties (measured by the McLachlan substitution matrix) of residues in catalytic sites, for 101 well-annotated enzyme su- vayinstitutes.be/ perfamilies. Superfamilies are ranked by similarity in catalytic residues. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/

Fig. 2. Superposition of two relatives from the CATH TPP superfamily. Pyruvate decarboxylase is in red and benzoylformate decarboxylase in pink with their catalytic residues shown in black and grey. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Convergent evolution of functions in superfamilies In some superfamilies, similar chemistries have been enabled by multiple evolution-

ary routes. Figure 2 shows homologous proteins which perform the same reaction but differ in their catalytic machinery. In the RuBisCo family, most plant species contain the C3 form, which is rather inefficient. Convergent evolution to a more efficient C4 form has occurred multi- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 19

What can Comparative Genomics Reveal About Protein Evolution? 19

ple times during evolution. By reconstructing the ancestral tree and building 3D models we found several stabilising mutations leading up to the transition between the two forms (C3/C4), whilst the transition itself is accompanied by signiﬁcantly

more destabilising mutations. By accumulating stabilising mutations the enzyme

can support the occurrence of destabilising mutations that have beneﬁcial impacts on the function. The transition to C4 is immediately followed by compensatory mutations which restore global stability.

Outlook to future developments distribution is allowed. distribution

An exciting new pool of sequence diversity and novel functional repertoires is emerg-

ing from the international metagenome studies. These initiatives are capturing sequence information from microbial communities found in a wide range of environ- 9180 ments (e.g., human gut, spoil, deep sea). Preliminary studies have already revealed

only. No further No only. many new relatives in known enzyme superfamilies and the tools developed by our

group (e.g., the CATH functional family proﬁles) and those of many other groups classifying protein superfamilies and functional families will shed light on the nature of changes in these proteins. As mentioned already the common active site exploited by relatives facilitates comparison of sequence changes in new relatives. vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS The challenge for the future will be to predict the nature of the shifts in function

- induced by these residue mutations. This will enable a better understanding and modelling of the active site environment and the mechanisms by which chemistries

http://www.sol can be modiﬁed and diverse substrates utilised.

host on on host Acknowledgments NLD thanks the MRC for funding. RST thanks NSF for funding. DL thanks NIH for funding and NF and SD thank the Wellcome Trust for funding. http://www.worldscientific.com/worldscibooks/10.1142/ References 1. B. H. Dessailly, R. Nair, L. Jaroszewski, J. E. Fajardo, A. Kouranov et al., Struc- ture 17, 869 (2009). 2. C. Orengo, J. Thornton, Annu. Rev. Biochem. 74, 867 (2005). 3. A. Cuﬀ, O. C. Redfern, L. Greene, I. Sillitoe, T. Lewis et al., Structure 17, 1051 (2009). ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 4. D. A. Lee, R. Rentzsch, C. Orengo, Nucl. Acids Res. 38, 720 (2010).

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 5. R. Rentzsch, C. Orengo, BMC Bioinform. 14, S5 (2013). 6. N. Furnham, I. Sillitoe, G. L. Holliday, A. L. Cuﬀ, R. A. Laskowski et al., PLoS Comput. Biol. 8, e1002403 (2012).

7. B. H. Dessailly, N. L. Dawson, K. Mizuguchi, C. A. Orengo, Biochim. Biophys. Acta 1834, 874 (2013). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 20

EXPLORING CHROMATIN BIOLOGY USING PROTEIN CHEMISTRY

TOM W. MUIR Department of Chemistry, Princeton University, Princeton, NJ, USA distribution is allowed. distribution

My view of the present state of research in protein chemistry 9180 Recent decades have witnessed astonishing progress in our ability to analyze native only. No further No only.

proteomes. Technologies centered on mass spectrometry have led to the recognition that proteins are decorated with post-translational modiﬁcations (PTMs) to an extent hitherto unimaginable. Indeed, it seems likely that every protein in a eukaryotic cell is modiﬁed, perhaps in manifold ways, at some point during its life-

vayinstitutes.be/ time. The number and type of PTMs found on proteins is ever expanding and PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

the resulting complexity of proteomes greatly complicates our ability to understand - cellular biology at the systems level. Perhaps nowhere is this complexity more evident than in chromatin, the nucleo-

http://www.sol protein complex that leads to spatial organization of genomic DNA in the eukaryotic nucleus. The fundamental repeating unit of chromatin is the nucleosome which com-

host on on host prises an octamer of four histone proteins (forming a proteinaceous spool) around which is wound the DNA. Histone proteins are modiﬁed in an astonishing number of ways; over 100 discrete PTMs have been reported at the time of writing this piece, and because histones form a hetero-octamer, the potential for combinatorial PTM

http://www.worldscientific.com/worldscibooks/10.1142/ signatures at the nucleosome level is enormous. Indeed, the last few years have seen a flood of genomic and epigenomic information that, among other things, reveal the existence of ‘chromatin states’ containing characteristic patterns of DNA and histone modifications that are correlated with different patterns of gene expression [1]. Such ‘cataloging’ is a critical first step towards an operational understanding of epigenetic control mechanisms, however, it does not in itself yield direct information on how these PTM patterns are installed, maintained and, most importantly,

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World what they are actually doing at the molecular level. For this, we still require more- directed biochemical and cell biological methods, which provide the quantitative NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY data needed to make testable statements on mechanism. Current biochemical approaches in the chromatin area are, however, very low throughput and so there is

an ever-growing disconnect between the ﬂood of “omics” data sets and our ability to follow up the numerous mechanistic hypotheses they generate. Thus, I believe that one of the grand challenges, and most urgent needs, in the chromatin area is the development of high-throughput protein biochemistry methods that will allow October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 21

Exploring Chromatin Biology using Protein Chemistry 21

us to acquire a more comprehensive molecular understanding of epigenetic control mechanisms.

My recent research contributions to protein chemistry

My research group has for many years focused on the development of methods that allow chemical access to proteins [2]. These protein semisynthesis approaches, along with complementary unnatural amino acid mutagenesis technologies [3], now make it possible to manipulate the entire covalent structure of proteins using the tools of synthetic organic chemistry. Among the very many things now possible, is the direct distribution is allowed. distribution generation of proteins bearing PTMs at user-deﬁned positions. Such proteins are

exceedingly hard to come by using classical biochemistry or molecular biology methods and, thus, represent a powerful application of these modern protein chemistries. 9180 For our part, we have largely focused on the histone proteins, which as noted earlier only. No further No only.

are elaborated with an astonishing range of PTMs. Over the last 4-5 years, we have put in place modular synthetic routes that allow the generation of histones carrying various PTMs, or combinations thereof, in biochemically useful quantities [1]. Modiﬁcations accessible to this approach include, methylation (at lysine and

vayinstitutes.be/ arginine), acetylation (lysine), ubiquitination (lysine), phosphorylation (histidine, PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

serine, threonine and tyrosine), sumoylation (lysine) and ADP-ribosylation (glu- - tamic acid). With these modiﬁed proteins as raw materials, we can, in principle, reconstitute any ‘chromatin state’ for detailed biochemical study. This reconstitu-

http://www.sol tion approach has begun to reveal the logic by which some of these modiﬁcations are installed (e.g., crosstalk mechanisms [4]) and readout (e.g., multi-valent interactions

host on on host [5]). Moreover, access to these reagents has yielded insight into how breakdown of epigenetic control mechanisms can lead to human disease [6]. Nonetheless, we have come to realize that our current modus operandi is too artisanal in Nature, meaning that it just takes too long to manufacture and test the chromatin complexes.

http://www.worldscientific.com/worldscibooks/10.1142/ The backlog of hypotheses we would wish to test, generated by the ongoing ‘omics’ initiatives, is simply overwhelming. I see this as the major bottleneck in the ﬁeld generally, assuming of course that one still buys into the notion that a mechanistic understanding of biochemical processes is a useful thing! Thus, I believe that the future in this area must fundamentally address this bottleneck, and that one productive approach to this would be to devise high-throughput protein biochemistry methods that accelerate the manufacture and functional analysis of these ‘designer

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World chromatin’ molecules and the factors that act on them. Achieving this goal will

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY have signiﬁcant challenges that will surely require innovative ideas from chemistry, biology and engineering.

Outlook to future developments of research on protein chemistry Remarkable progress has been made in the ﬁeld of protein chemistry in recent years. I think we have transitioned from asking whether or not a modiﬁed protein can be October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 22

22 T. W. Muir

made, to perhaps the more exciting question of what can we learn about biology given that we probably can make it (certainly true for modiﬁed histones). This is not to say that we can synthesize any protein imaginable, and certainly there

is still room for improved and streamlined methods. However, I would suggest

that in thinking about such innovations, we should do so with an eye towards the even bigger problem, namely the very large number of PTM-modified proteins found in proteomes and how to functionally characterize them all. It seems likely to me that tackling this problem will have to involve the parallel manufacture of hundreds, perhaps thousands, of chemically defined, modified proteins and that this process will also have to incorporate a solution to the issue of keeping track of distribution is allowed. distribution all these molecules thereby allowing one to perform massively parallel quantitative

biochemistry. This is a complex problem, but one that I think represents the next chapter in the ﬁeld of protein chemistry. 9180 only. No further No only. Acknowledgments I am grateful to the US National Institutes of Health for supporting the research in my laboratory. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- References 1. B. Fierz, T. W. Muir, Nat. Chem. Biol. 8, 417 (2012).

http://www.sol 2. M. Vila-Perell´o, T. W. Muir, Cell 143, 191 (2010). 3.L.Wang,J.Xie,P.G.Schultz,Annu. Rev. Biophys. Biomol. Struct. 35, 225

host on on host (2006). 4. R. K. McGinty, J. Kim, C. Chatterjee, R. G. Roeder, T. W. Muir, Nature 453, 812 (2008). 5. A. J. Ruthenburg, H. Li, T. A. Milne, S. Dewell, R. K. McGinty et al., Cell 145, 692 (2011). http://www.worldscientific.com/worldscibooks/10.1142/ 6. P. W. Lewis, M. M. M¨uller, M. S. Koletsky, F. Cordero, S. Lin et al., Science 340, 857 (2013). ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 23

OUR EXPANDING PROTEIN UNIVERSE

ADAM GODZIK Program in Bioinformatics and Systems Biology, Sanford-Burnham Medical Research Institute 10901 N. Torrey Pines Rd, La Jolla, CA 92037, USA distribution is allowed. distribution

My view of the present state of research on new chemistry in the expanding protein universe 9180 Last few years witnessed an unprecedented explosion of information on protein only. No further No only. coding genes from individual species (genomics) and environments (metagenomics), many of them coding for enzymes with unknown functions and speciﬁcities. We have now amassed sequences of billions of protein coding genes from over 200 eukaryotes, thousands of bacteria and archea as well as from microbial communi-

vayinstitutes.be/ ties (microbiomes) from thousands of environments, sampling from a vast “protein PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- universe” of all existing protein [1]. Actually, over 99% of all known proteins are coming from the latter group and for the most part are not associated with specific organisms nor with specific neighborhoods on their genomes. This makes the func- http://www.sol tion prediction for them very difficult. Based on ratios of different types of proteins in bacterial genomes, we can estimate that between 30 and 40% of these genes code host on on host for enzymes, for some of which we can predict general functions (hydrolases, proteases, kinases, etc.) but as much as 40% remain completely unknown. We can only speculate that many of them would be involved in metabolism of various natural compounds. http://www.worldscientific.com/worldscibooks/10.1142/

My recent research contributions to new chemistry in the expanding protein universe Over last several years in our group we have collected and analyzed genomic and metagenomic data sets from diﬀerent sources. Public protein databases, such as translations of the GeneBank [2] coding sequences and UniProt [3], collecting pro-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World tein sequences from genomic studies and sequencing of individual genes, undergo a steady, exponential growth (see Fig. 1, blue rombi) and toward the end of 2013 their NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY size approached over 45M non-redundant protein sequences (http://uniprot.org). However, in last few years, the biggest impact on the growth of protein sequence

databases came from metagenomic studies, where DNA sequencing is applied directly to microbiomes, complex microbial communities that occupy almost every imaginable environmental niche on Earth. By skipping the traditional step of growing microbes in pure cultures, we are now able to sample DNA of organisms that October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 24

24 A. Godzik

distribution is allowed. distribution Fig. 1. Growth of protein sequence datasets, UniProt TrEMBL (blue rhombi) and JGI IM/G 6 (red squares). The Y axis scale is in million (10 ) sequences. 9180 only. No further No only.

don’t grow in such cultures, at least not easily. Recently, developments in the study of a human metagenome, dubbed by some a second human genome [4], garnered a lot of attention. With several large studies around the world, from NIH Human Microbiome Project [5, 6] to European/Chinese metaHIT project [7], already iden-

vayinstitutes.be/ tiﬁed 100M of predicted protein sequences, doubling the size of traditional datasets. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

However, studies of environmental microbiomes, such as present in soil and marine - environments dwarf all previous eﬀorts, providing to date over 16B (109) predicted protein sequences (Fig. 1, red squares) in only one of several metagenomic resources,

http://www.sol IMG/M at the DOE Joint Genomics Institute [8]. Our protein universe is now over 17B large! Even if expected redundancies and wrong ORF calls currently inherent

host on on host in metagenomics data would reduce this number, the pace of discovery of new genes would not stop and it is clear that early estimates of the size of the protein universe were incorrect and a much larger part of protein space of all possible protein sequences is actually explored in Nature.

http://www.worldscientific.com/worldscibooks/10.1142/ These developments bring the question of what it the internal structure of the newly discovered part of protein universe. The part studied so far, for the most part can be classified into a hierarchical system of protein families and superfamilies. In- formation about such families is collected in databases such as Pfam [9], SMART [10] or CCD [11] with the number of defined protein families counted in tens of thousands, for instance there are 14,831 protein families defined in the most recent Pfam release (27.0). As shown in Fig. 2 (lowest panel), over 85% of proteins in a

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World set of 220 fully sequenced genomes representing all phyla with sequenced representatives can be matched to a Pfam family. This coverage drops to little over 60% for NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY metagenomics sets from the human metagenome focused metaHIT [7] and type 2 diabetes studies [12] (top and middle panels in Fig. 2, respectively). Similar level

of coverage is seen in other metagenomics datasets (data not shown), suggesting that while the diversity and novelty of genes found in these microbial communities is signiﬁcant, the core is formed by known and deﬁned protein families. At the same time, its worth noting that even well studied microbial genomes still contain October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 25

Our Expanding Protein Universe 25

"#$%&'$ 1#2 SW +*,

27.# T "' #2'#1 1#2 SS +*,

distribution is allowed. distribution TS[ 0#.0#1#,2 2'4# !2#0' * %#,-+#1

RTXV +*,

9180 R$ TR$ VR$XR$ZR$ SRR$

only. No further No only. $ -0$1 !-4#0#" 7 .$ + $ +'*'#1

$ -0$1 ,-2 !-4#0#" 7 .$ + $ +'*'#1

Fig. 2. Coverage of diﬀerent subsets of protein space with currently deﬁned protein families. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- a significant number, around 15%, of proteins that cannot be classified into families using current resources. Another comment here is that classification of a protein

http://www.sol into a known protein family is not synonymous to assigning function to this protein, as many protein families in Pfam and other similar databases do not have any

host on on host veriﬁed functional annotations and thus are known as domains of unknown function (DUF) families. Since ﬁrst metagenomic datasets became available, our group focused on identifying novel protein families in these datasets, with special emphasis on families

http://www.worldscientific.com/worldscibooks/10.1142/ of secreted proteins from microbes present in the human gut microbiome [13]. As of today, close to 400 such families have been identified and deposited in the Pfam database. In collaboration with the Joint Center for Structural Genomics, an NIH Protein Structure Initiative center for high-throughput structural determination, we have systematically targeted these families for structure determination, solving structures of first representatives of over 40 of these families. However limited in scope to a specific segment of protein universe, this explo-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World ration brought rather unexpected results that give us an interesting perspective of what can we expect from the future, more extensive studies of novel proteins from NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY new regions of the expanding protein universe. Most if not all families identiﬁed in the human gut microbiome represented divergent branches of already known

families, with diﬀerences on the sequence level that often made it impossible to recognize the homology, but with structural similarities so strong that the couldn’t be explained by convergent evolution. While functions of protein in many of these new families remain unknown, we have identiﬁed novel families of glycosyl hydro- October 14, 2014 16:35 WSPC Proceedings - 9.75in x 6.5in master/pg 26

26 A. Godzik

lases and other carbohydrate metabolic enzymes [14, 15] — as expected in the gut environment, rich in polysaccharide based plant ﬁbers such as starch, and multiple families of proteins involved in cell-cell interactions [16, 17], again expected in the

dense environment of the gut.

Outlook to future developments of research on new chemistry in the expanding protein universe We can expect even faster growth of protein sequence databases and explorations of novel environments, resulting in the known protein universe becoming even big-

distribution is allowed. distribution ger and more diverse. With the help of other high-throughput techniques such a

proteomics and metabolomics we expect that integration of these multiple sources

of information would be both a main challenge, but also a main source of discov- 9180 ery of new functions, new chemistry and new processes in the new environmental niches. The expanding protein universe is going to catch up with equally extensive only. No further No only.

world of natural products, which are all after all synthetized and metabolized by protein-based catalysis.

vayinstitutes.be/ Acknowledgment PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Research described here was funded by the NIH grant U54 GM094586 (JCSG).

References http://www.sol 1. I. Ladunga, J. Mol. Evol. 34, 358 (1992).

host on on host 2. D. A. Benson, M. Cavanaugh, K. Clark, I. Karsch-Mizrachi, D. J. Lipman et al., Nucl. Acids Res. 41, D36 (2013). 3. The UniProt Consortium, Nucl. Acids Res. 41, D43 (2013). 4. E. A. Grice, J. A. Segre, Annu. Rev. Genom. Hum. Genet. 13, 151 (2012).

http://www.worldscientific.com/worldscibooks/10.1142/ 5. L. M. Proctor, Cell Host Microbe 10, 287 (2011). 6. P. J. Turnbaugh, R. E. Ley, M. Hamady, C. M. Fraser-Liggett, R. Knight et al., Nature 449, 804 (2007). 7. J. Qin, R. Li, J. Raes, M. Arumugam, K. S. Burgdorf et al., Nature 464,59 (2010). 8.V.M.Markowitz,I.M.Chen,K.Chu,E.Szeto,K.Palaniappanet al., Nucl. Acids Res. 40, D123 (2012).

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 9. M. Punta, P. C. Coggill, R. Y. Eberhardt, J. Mistry, J. Tate et al., Nucl. Acids Res. 40, D290 (2012). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 10. I. Letunic, T. Doerks, P. Bork, Nucl. Acids Res. 40, D302 (2012). 11. A. Marchler-Bauer, S. Lu, J. B. Anderson, F. Chitsaz, M. K. Derbyshire et al.,

Nucl. Acids Res. 39, D225 (2011). 12. J. Qin, Y. Li, Z. Cai, S. Li, J. Zhu et al., Nature 490, 55 (2012). 13. K. Ellrott, L. Jaroszewski, W. Li, J. C. Wooley, A. Godzik et al., PLoS Comput. Biol. 6, e1000798 (2010). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 27

Our Expanding Protein Universe 27

14. D. Das, M. Herva, M. A. Elsliger, R. U. Kadam, J. C. Grant et al., J. Bacteriol. 195, 5555 (2013). 15. A. P. Yeh, P. Abdubek, T. Astakhova, H. L. Axelrod, C. Bakolitsa et al., Acta

Crystallogr. Sect. F Struct. Biol. Cryst. Commun. 66, 1287 (2010).

16. Q. Xu, B. Christen, H. J. Chiu, L. Jaroszewski, H. E. Klock et al., Mol. Micro- biol. 83, 712 (2012). 17.Q.Xu,P.Abdubek,T.Astakhova,H.L.Axelrod,C.Bakolitsaet al., Acta Crystallogr. Sect. F Struct. Biol. Cryst. Commun. 66, 1281 (2010). distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 28

THE SCIENTIFIC IMPACT OF FREELY AVAILABLE CHEMICAL PROBES

ALED M. EDWARDS Structural Genomics Consortium, University of Toronto 101 College Street, MaRS South Tower, Suite 706 Toronto, ON M5G 1L7, Canada distribution is allowed. distribution

9180 Enabling research eﬀorts on the proteome The human genome sequence provided a near complete list of human genes and pro- only. No further No only. teins. A comprehensive understanding of human normal and pathobiology will come only after the role(s) of each human gene and protein, and their inter-relationships, are well understood. However, the anticipated impact of having our genome sequence is not being quickly realized. Bibliometric analysis of research activity on vayinstitutes.be/ each human gene/protein revealed a clustering of activity on a relatively small sub- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- set of human genes/proteins, highly enriched in those that had been the subject of research eﬀorts for decades [1, 2]. Progress toward a deeper understanding of human

http://www.sol biology and the discovery of new medicines is being limited by this behaviour. Additional bibliometric analysis also pointed to a partial solution. Historically,

host on on host the availability of high quality, small molecule inhibitors of proteins provided impe- tus for the research community to initiate research on proteins that previously had garnered little attention, provided the small molecule inhibitor was made available without restriction on use. This raised the hypothesis that research activity could be increased on proteins residing among the less studied by making chemical research http://www.worldscientific.com/worldscibooks/10.1142/ tools readily available. To test this hypothesis, we have engaged in a large-scale collaboration with academic and pharmaceutical medicinal chemists with the aim to produce, and make available, cell-active small molecule inhibitors of proteins and enzymes that regulate chromatin biology.

Our recent research contributions

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Our project is focused on families of human proteins that regulate protein acetylation and methylation. Speciﬁcally our objective is to generate inhibitors of enzymes NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY that acetylate and methylate other proteins, enzymes that demethylate methylated proteins, and the suite of proteins that bind preferentially to acetylated or methy-

lated side chains [3, 4]. These modiﬁcations regulate, among things, proteins that control gene expression. The aim of the project is to increase research activity on these protein families, and in so doing, to deﬁne new links between their members and human disease. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 29

The Scientiﬁc Impact of Freely Available Chemical Probes 29

The project revolves around the technology platform within the Structural Ge- nomics Consortium (SGC). Within the aforementioned protein families, for most family members, the SGC has generated puriﬁed proteins, their 3D structures and

has built suites of biophysical and biochemical assays to monitor protein function.

This core resource enables a large number of medicinal chemistry groups to identify weak inhibitors and, using structure as a guide, to advance these to become potent, selective and cell-active molecules. With nine pharmaceutical companies and a network of academics involved, the project has produced “chemical probes” for a number of proteins, some of which have been used to uncover new biological insights [5–9]. Importantly, the more distribution is allowed. distribution general inﬂuences on academic research activity are beginning to appear. The

papers reporting the chemical probes for the G9a methyltransferase [6], the JMJD3 histone demethylase [7] and the BET family of bromodomains [5] have become the 9180 top cited papers on those proteins since their publication (Edwards, unpublished). only. No further No only. In turn, the research activity on these proteins, compared to other family members, has increased signiﬁcantly (Fig. 1). With these early indications in hand, it does appear that the availability of chemical probes is suﬃcient to induce the research community to focus on “novel” proteins. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/

Fig. 1. Publication trends on bromodomains-containing proteins. Annual publications naming bromodomains-containing proteins in the title, abstract or key words were quantiﬁed using the ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Scopus database (blue line). Papers referencing the BET subfamily of bromodomains are indicated in the red line. The black arrow indicates when the BET chemical probe papers were published. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Outlook to future developments

The availability of the human genome sequence places biomedical research in a unique situation among all sciences — it provides us a map of the known unknowns. However, even given this map, the research funding system appears fearful to go October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 30

30 A. M. Edwards

exploring. Although there is strong justification to continue studying proteins of known importance or biomedical interest, there is equal justification to encourage more research into the unknown. But this is difficult to accomplish within the

current global research funding framework, in which funding is allocated along na-

tional lines and according to the opinions of a group of peers. It is my opinion that scientists would be willing explorers if the granting, tenure and peer-recognition systems permitted. But of course, these systems were created and are endorsed by the research community — we have met the enemy and it is us.

Acknowledgments distribution is allowed. distribution

The Structural Genomics Consortium is a registered charity (number 1097737) that receives funds from AbbVie, Boehringer Ingelheim, Genome Canada, GlaxoSmith- 9180 Kline, Eli Lilly Canada, Janssen, the Novartis Research Foundation, Pﬁzer, Takeda,

only. No further No only. the Canada Foundation for Innovation, the Ontario Ministry of Economic Develop-

ment and Innovation, and the Wellcome Trust.

References

vayinstitutes.be/ 1. A. M. Edwards, R. Isserlin, G. D. Bader, S. V. Frye, T. M. Willson et al., Nature PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 470, 163 (2011). 2. http://baderlab.org/Data/RoadsNotTaken

http://www.sol 3. D. J. Patel, Z. Wang, Annu. Rev. Biochem. 82, 81 (2013). 4.C.H.Arrowsmith,C.Bountra,P.V.Fish,K.Lee,M.Schapira,Nat. Rev. Drug

host on on host Discov. 11, 384 (2012). 5. P. Filippakopoulos, J. Qi, S. Picaud, Y. Shen, W. B. Smith et al., Nature 468, 1067 (2010). 6. M. Vedadi, D. Barsyte-Lovejoy, F. Liu, S. Rival-Gervier, A. Allali-Hassani et al., Nat. Chem. Biol. 7, 566 (2011). http://www.worldscientific.com/worldscibooks/10.1142/ 7. L. Kruidenier, C. W. Chung, Z. Cheng, J. Liddle, K. Che et al., Nature 488, 404 (2012). 8.W.Yu,E.J.Chory,A.K.Wernimont,W.Tempel,A.Scoptonet al., Nat. Commun. 3, 1288 (2012). 9. L. I. James, D. Barsyte-Lovejoy, N. Zhong, L. Krichevsky, V. K. Korboukh et al., Nat. Chem. Biol. 9, 184 (2013). ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 31

SESSION 1: NEW CHEMISTRY IN THE EXPANDING PROTEIN UNIVERSE

CHAIR: CHRISTOPHER T. WALSH AUDITORS:K.BARTIK1,E.GOORMAGHTIGH2 (1) Engineering of Molecular Nanosystems, Universit´e Libre de Bruxelles, CP 165/64, 50 av. F.D. Roosevelt, 1050 Brussels, Belgium distribution is allowed. distribution (2) S.F.M.B., Universit´e Libre de Bruxelles, CP206/2, Campus Plaine, Bd du Triomphe,

B-1050 Brussels, Belgium 9180 only. No further No only. Discussion among panel members Christopher Walsh to Aled Edwards: Aled, it’s not as if these molecules are pheromones for humans. How did you decide what probes to make? Suppose I don’t care about kinases and methyltransferases from all domains? You are still vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS looking under a lamppost in a sense.

Aled Edwards: Absolutely! For us, it was pragmatic. If you work on protein

http://www.sol families, the chemistry is shared and you don’t have to invest so much in chemistry because you have common libraries that provide starting points. I think we just

host on on host had to try and show that the model works ﬁrst and then hopefully it will be copied.

Aled Edwards to Adam Godzik and Christine Orengo: You have been at these meetings for a decade and everyone is talking about the domains of unknown function. To your knowledge, has there been any concerted eﬀort around the world, http://www.worldscientific.com/worldscibooks/10.1142/ anywhere, to say, “let’s start looking at these” because the probability of ﬁnding out interesting or novel science is higher there? To my knowledge, there is none!

Adam Godzik: Surprisingly, as far as I know, nobody other than us, nobody has picked it up, but it’s certainly coming. What we noticed is that, when you publish papers or describe them, in time, after 2 or 3 years, there is a peak of interest. For us, the catalyst of interest is publishing papers on them. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Christine Orengo: I think the problem is, as you say, that people are quite nervous NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY about studying proteins for which there is no information at all and I suppose that’s why we try to organize things so that we can group the novel proteins so that at

least we can say what they are most similar to. So there are some clues there as to what their functions might be, or where they might have diverged from. You then have a starting point and a few clues. I think that if you are talking about some of these completely novel families we actually have no insight at all and you are talking about a lot of experimental work. You have got to do a vast amount of screening to October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 32

32 New Chemistry in the Expanding Protein Universe

get any idea what they could be doing. So, I think at the moment the way people are using this data, is to take sequences that have at least some similarities to a family for which something is known about the active site and see what the changes

might be. From that, some deductions can be made about which ligands might

bind, and then they’ve got a clue. In this way it is not such an expensive operation. I think what you’re proposing is very exciting but it is also hugely expensive.

Wilfred van der Donk to Aled Edwards: I was a little bit surprised that RNAi didn’t seem to have made much of a dent. I would have thought that it would level the playing ﬁeld much more than it seems to have. distribution is allowed. distribution

Aled Edwards: This is an interesting observation. When you get the RNAi knocked

down, if you do it en masse, then you have to think “What do I do next?” I need

9180 a cell line, I need a reagent to prove that the protein was knocked down, I need a whole bunch of stuff that costs me, or my students, a lot of time and effort to get to only. No further No only. the point where I can actually do an experiment. And if we have a five-year grant or a three-year PhD student, often it is proving not enough.

Jean-Louis Reymond to Aled Edwards: Counting papers is one thing but you have

vayinstitutes.be/ to measure the depth of insight, so I think it’s only natural that there are diﬀerent PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- levels of insight. The question is: “Have you measured whether the additional papers add anything signiﬁcant or can one measure the learning level of an additional publication?” http://www.sol

Aled Edwards: It’s a hard experiment but you would imagine that, because we host on on host continue to publish in those areas, because the funding system now is quite austere and only good, great grants get funded, that a lot of the papers on the old chestnut proteins are important. It’s not that they are not important proteins, it’s just that as a scientiﬁc institution, maybe we don’t have the balance right between them.

http://www.worldscientific.com/worldscibooks/10.1142/ You can discover interesting things about these proteins. As Tom said, they are all probably modiﬁed by weird and wonderful things and they all have interesting activities, and so it’s not as if the science is poor or is yesterday’s science, it is just that we tend to focus on what we have always done. As Roger Kornberg said, we like to “fondle our problems”.

Christine Orengo to Adam Godzik: You’re analysing in much more detail some of

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World these very novel families and as you say, it is something we ﬁnd as well, that quite often they turn out to be remote homologues of what we have seen already. And NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY from our sort of preliminary scans of the meta-genome data, we see that a lot of the new enzymes that we pick up are still relatives of some of the very common

favourable structural scaﬀolds like the TIM barrel or the Rossmann fold. We see them over and over again appearing in this dataset and I just wondered, because you have done a larger scan, whether you see some other types of structural scaﬀolds also emerging and being exploited quite frequently by these new sequences you are analysing? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 33

Session 1: Discussions 33

Adam Godzik: Yes, we did and the result was sort of depressing in the sense it looks like the most popular folds are getting even more popular. So, in answer to my question on whether this new environment would exploit these new scaﬀolds,

they don’t. They actually increase the popularity of the most popular fold. So yes,

in genomes TIM barrels represent 10%, in meta-genomes, they represent 15% and so the distribution is getting even more skewed.

Tom Muir to Chris Walsh and Wilfred van der Donk: You spoke about these amazing peptides and natural products. I wondered though where we are in terms of understanding of what all these things are doing. Do we even know in many distribution is allowed. distribution

cases which organisms they are targeting in the ecology of the soil or wherever they

come from? Does informatics have any role to play in predicting any of this stuﬀ?

9180 Wilfred van der Donk: In most cases, we don’t. As you pointed out, we know which organism produces them in most cases but we don’t necessarily know enough only. No further No only. about the ecology, where they’re living, who they are targeting in their natural environment. We can test them against a panel of organisms that we are interested in as humans, and that’s what most people do. Most people don’t really test them against the organisms they are probably living with. Can we predict activities? To vayinstitutes.be/ some extent, but I think to a low-level, you can sometimes say “well this looks like PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- something we have seen before” and we can predict the activity but as soon as it goes away from something that we know and have seen before, we essentially have

http://www.sol no idea. And I would say this is actually a problem for genome mining because a lot of the molecules we have made by genome mining. When we have the molecule we

host on on host can see that indeed it is the molecule the organism also produces. After we do some tricks to try have it produce the compound, then we see no bioactivity whatsoever. And we have no idea really to predict what clusters to go after! And hence, as you said in your talk about your ﬁeld, we have to do this in scale in order to really get back to where we were in the 60’s and 70’s where simply by a large number of http://www.worldscientific.com/worldscibooks/10.1142/ natural products being available a number of them made it to pharmaceuticals.

Tom Muir: How much interaction is there between chemists, such as yourself, and microbiologists or ecologists who actually are thinking about bacterial communities in the soil, in the ocean or in our gut? Do you guys talk to each other to try to think about a concept?

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Wilfred van der Donk: Yes, I collaborate with several microbiologists both at Illinois

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY and elsewhere to talk about the chemical ecology. I am mostly interested in the chemistry, but yes we have to ﬁnd out what these molecules are doing. As you probably know, there is even an hypothesis out there, that what we call “antibiotics” are actually not made in the environment to kill other organisms but are more signalling molecules. And so yes, we do. I don’t know if Chris wants to add something to both of your questions? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 34

34 New Chemistry in the Expanding Protein Universe

Chris Walsh: First I would make a distinction between eukaryotes, prokaryotes and multicellular eukaryotes where I think there isn’t much interesting new chemistry that has been discovered. That may be a red ﬂag for somebody like Aled Edwards.

Are nuclear hormone receptors actually interesting from a point of view of new

chemistry? I’ll come back to that. It is meant to be an interesting but not an- tagonistic statement. Let me go to fungi which are a fascinating set of organisms which make many of these unusual molecules and almost nothing is known about the natural physiology of what these molecules do. I suspect part of that is the historical view that you can ﬁgure out the function of conditional metabolites in laboratory-based cultures. Almost clearly the missing links are the fact that you distribution is allowed. distribution have to look at ecologic communities. It has been estimated that in the kinds of

molecules I or Wilfred or others study, maybe there are 30 gene clusters in the typical fungal cells or Actinomyces and only 5 to 10% of them have been turned 9180 on, transcribed under laboratory conditions. So the other 90% of the conditional only. No further No only. metabolite pools are just not being exploited in the physiology that we look at. To me, it is about looking under the lamppost and that’s what our community does. I’m hard pressed to ﬁnd a paper for example on kinases which really is interesting or novel in terms of chemistry, that is a diﬀerent question from the health relevance

vayinstitutes.be/ and the applied importance. It is worth factoring those two things though. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Aled Edwards to Adam Godzik: I have a question about anaerobic and aerobic biology. If you parse your datasets by strict anaerobes and aerobes, do you see a

http://www.sol diﬀerence in the sort of patterns of protein family that are in each?

host on on host Adam Godzik: Yes we do. We haven’t really followed on that, but yes there are definitely diﬀerences. We even have a theory that the distribution of folds in anaerobes is sort of more ancient because they map to families, which form certain speciﬁc parts of the metabolic network, which people argue are older. So we thought we saw a pattern in this distribution. http://www.worldscientific.com/worldscibooks/10.1142/ Aled Edwards to Chris Walsh: The anaerobic biochemistry is really hard, you’ve got to go under hoods and they have cofactors. . .

Chris Walsh: It’s not hard for the anaerobes!

Aled Edwards: Exactly, and they have cofactors, diﬀerent corrinoids. So if it’s a

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World source of great chemistry, where are we going to get the biochemists that can do it because we can probably count on one hand the people who are really skilled in the NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY art of anaerobic biochemistry.

Chris Walsh: I think that in the general discussion later on and in the session this afternoon, we’ll hear about the “enzyme function initiative” people and they have just built the anaerobic biochemistry in line with several stations from, doing assays, to doing puriﬁcations, to large throughput crystallography. So it will take October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 35

Session 1: Discussions 35

those kinds of activities to reduce the barriers. So it’s not just “Are chemicals available?” but also getting through the technology. Just a comment — one of my ﬁrst comments on my ﬁrst slide was that I thought anaerobes were interesting

and I’ll just give you one example: the reduction of benzene to cyclohexane can

be done by anaerobes but not by aerobes. It’s a sort of interesting Birch type reduction chemistry. As far as I know, reported in the literature 25 years ago but never studied in any detail.

Wilfred van der Donk: I would like to add to the discussion. Much of metallo- biochemistry requires the same type of very controlled environment. There is actu- distribution is allowed. distribution

ally a really large research community that is set up to study these enzymes. So,

whether it is from an anaerobe or an aerobe, you often run into oxygen-sensitive enzymes and so there are a fairly large number of people that are set up to do that. 9180 Aled Edwards: Yes, some of the problems though are that you cannot make them only. No further No only. recombinantly because the hosts sometimes don’t have the appropriate cofactors that their host organisms would make. So you need to grow them, but they double every two weeks and you just have to wait. It’s hard to do bucket-biochemistry on a thing that is barely alive as you know. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Christine Orengo to Adam Godzik: My question is about the orphan families that he’s ﬁnding, whether he has done any further analysis on the sort of characteristics they have? Some studies we have done suggest that they’re perhaps more disor- http://www.sol dered, or have higher proportions of disorder, so it’s harder to spot any relationship with other known families. Some of the relationships may have been obscured. So host on on host I just wondered, have you looked at the features of those orphan families? Perhaps they might be less tractable for structural determination as well?

Adam Godzik: No, we didn’t do it in this way. What we were interested in is

http://www.worldscientific.com/worldscibooks/10.1142/ to see how quickly they can be de-orphanized. So what we noticed is that calling something an orphan, meaning that has no homologue, is very time-dependent because most of the orphans known at a given time would ﬁnd homologues once the dataset gets larger. But then more orphans would be introduced so it’s like you’re doing a cycle. So in some sense, them being orphans is nothing special about them. We just happened to see the top of the mountain ﬁrst and then we looked deeper and we saw the rest. Our thought is that there is nothing special about them — it

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World is just random events. These were not very large families and we just found just one

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY representative, we looked again and we see more and more. They are not special anyway. By random chance, we found one example ﬁrst and then, in next dataset, we found plenty more.

Christine Orengo: So do you think it’s reasonable to suppose that, as the data accumulate the information we have for each family grows and actually the sequence patterns we derive become much more sensitive and so we can recognize these very October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 36

36 New Chemistry in the Expanding Protein Universe

distinct homologues? Rather than the universe exploding if you like, actually it will start to collapse again. We will be able to bring these relatives back into their correct superfamily. At that point, what becomes interesting is the ways in

which the diﬀerent domains combine because that seems to me where you do have

a huge explosion of data that is going to be much harder to understand. At that point, you have to try to understand what the individual domain is contributing to the function of the multi-domain assembly. That’s a diﬀerent problem that we have barely started to tackle. Once we have organized the sequences into their constituent domains we have the central building blocks of these proteins. Then we have to start to understand how, when you combined them in diﬀerent ways, you’re distribution is allowed. distribution again starting to modify their functions. I think that is something we have barely

started doing.

9180 Adam Godzik: Two questions in one. We tried to do both. The ﬁrst one we

only. No further No only. believed it would happen, that we would start to saturate at some point and that

we would become so good in recognizing families that all these orphans would be “folded back”. It doesn’t seem to be happening yet. Yes it happens for some orphans but then for every orphan we can add to a family we have 10,000 more coming. We are still in a completely linear growth curve in terms of orphans. So vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS we can all speculate when it will happen but we haven’t yet seen the traces of

- such rating. For the second question, I agree with you. We tried to work on this too but I think this is much more important for eukaryotes. Most bacterial and

http://www.sol microbial proteins are single domain and we don’t have so many problems or so many issues understanding how domains contribute, unless you go into operons.

host on on host The diversity in eukaryotes is mostly built by reshuffling already known domains. We definitely have two very different methods of how, bacteria or microbes versus multicellular eukaryotes, deal with a complex world. One of them is diversifying into the sequence space and the designing of more and more homologues, the others are actually shrinking their existing domains by recombining them into new patterns. http://www.worldscientific.com/worldscibooks/10.1142/ It’s definitely different paths and I think you mentioned that in yeasts, for instance, there are definitely different patterns of diversification, of finding new chemicals they need or new challenges by mixing domains rather than inventing new ones.

Chris Walsh: I want to go back to Adam’s comment. So far, it looks like there are 1300 or so protein folds and you estimate it might be possible to have 5000 by, I think you said graph theory or perhaps I misunderstood. At what point does it ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World not become worthwhile looking for new folds, that it’s a diminishing return and one NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY should be doing something else? Or is that just coming along for the ride and it’s not a frontier problem anymore?

Adam Godzik: We are not looking for new folds speciﬁcally. We cannot really predict whether the protein will have a new fold. But we’re looking for new families and increasingly we’re ﬁnding that even if they look novel, they have an old fold. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 37

Session 1: Discussions 37

The fact that the number of folds seems to be saturating is a result, it is not an assumption. It’s surprising because we thought it would keep growing and it is deﬁnitely on the level of saturation. We very rarely ﬁnd new folds.

Christine Orengo: If I can contribute to that. Yes in fact I think that the number

of folds is reducing as we get more information. I think it is rather difficult how to define a fold. It’s a quite controversial discussion about what we mean by fold! If you think about the core of proteins, the core structural scaffold, I think really there only seems to be a thousand of those arrangements seen over and over again. During the 20 years that we have been classifying protein sequences and structures distribution is allowed. distribution

into families, the number of folds has actually started to diminish as we deﬁne these

core structures more clearly. I completely agree with Adam and I don’t think we are likely to see many more new folds. It’s the variation, it’s the sort of peripheral 9180 decoration to these folds that are interesting and we are going to see many more of

only. No further No only. those.

John Gerlt to Chris Walsh: I think every time Chris Walsh says something I have to add a slide to my talk this afternoon. I will make a comment. I was at a workshop a couple of years ago with DOE and, Claire Frasier who was at the workshop, made vayinstitutes.be/ the comment that maybe what we should do is stop sequencing and maybe ﬁgure PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- out exactly how we decipher what’s already in the database. That is, how do we go about ﬁguring out what the functions are of the proteins that we have now? The

http://www.sol problem is only getting worse, and I don’t think Claire was terribly serious about stopping sequencing, but the problem is overwhelming. I think, and I’m going to

host on on host talk about this a bit this afternoon, that one needs not just bioinformatics (I don’t mean to insult bioinformatics) but really a multidisciplinary approach to ﬁgure out what all this sequence information means. Is there some way to decipher function from the information? I think that’s the real problem, from my perspective, that is confronting this community. If we want to discover new chemistry, we are going http://www.worldscientific.com/worldscibooks/10.1142/ to have at some point to stop and ﬁgure out what that chemistry is and how do we go about doing that.

Chris Walsh to Tom Muir: You said in passing Tom, regarding histone modifications in chromatin, that essentially every combination of acetylation, methylation, phosphorylation, ubiquitylation on a residue that could be modified, can be seen. Maybe I’m putting words in your mouth. What is the chance that those are random and ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World not useful? Could they represent promiscuous activity of post-translational modifi-

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY cations enzyme sets and one gets a statistical or a bio-statistical set? And some of the combinations which you or others may make are not going to be physiologically relevant. How is one ever going to sort that out? And I pick that because you

pointed out there are so many combinations, many of which you’ve actually seen, in contrast to the antibiotic case that van der Donk mentioned where you only see a very small fraction. What is the chance those are either unplanned, not purposeful, October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 38

38 New Chemistry in the Expanding Protein Universe

mistakes or maybe they are evolving from new functions? What do you think about that?

Tom Muir: It is a great question. I guess to rephrase: how much collateral damage is really on these proteins, very abundant proteins? To answer your question in

short we really don’t know. . .

Chris Walsh: Maybe they’re just things for acylating and methylating chemically active side chains in proteins. . .

Tom Muir: I mean essentially anything that can be put on CoA is now being distribution is allowed. distribution found on histone tails. So how much of that really has function versus it just being collateral damage? We really have no idea. This is the problem. The more sensitive

9180 the analytical techniques become (to dig into the noise), the more of these things you will ﬁnd and so, there lies the challenge. . . only. No further No only.

Chris Walsh: That’s the curse of modern mass spectrometry!

Tom Muir: That’s actually true but it doesn’t mean that it is not incredibly interesting of course! But one has to think long and hard before one undertakes a vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

2-year expedition to make these things and try to study their biochemistry. If you’re - wrong, if it really is noise, which I see as being one of the biggest problem here, we need ways to speed this up because there is going to be noise some of the times.

http://www.sol We have to rethink how we approach this problem.

host on on host Aled Edwards to all chemists: I went to a plant meta-genome meeting and there are two things I want to mention. The ﬁrst, they asked, “Why aren’t there any soil microbiologists”? They ﬁgure that they don’t chat with one another at this meeting anyway which was kind of interesting. And there are no plant people here I notice, I don’t think. There is a lot of chemistry in plants you know. The second http://www.worldscientific.com/worldscibooks/10.1142/ thing is that most of the talks in the plant meeting were on carbon and nitrogen, and trace metals are essential for microbes. No one worked on halogen cycles, trace metals. Is there a chemical myopia to the kinases where oxygen and sulphur, and carbon, and nitrogen are on this side, and equally important for life molecules are understudied? I mean I don’t know the literature but it sounds like there is.

Wilfred van der Donk: I think you are right and I think that, at least in the area ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World of natural products, it is obvious why plants are so much harder, why they are not

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY being looked at. The genes are not clustered and so you don’t get these very nice packages of genes that you can look at for chemistry. In respect to metals in plants, I’m not qualiﬁed to comment as to why. First of all, whether people do or do not

study them, of course things like the photosystem is very well studied. But you could argue that it is easier studied in bacteria even, then in plants. But yet, I think for plants, as you pointed out in your answer to my question about RNAi, October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 39

Session 1: Discussions 39

theoverheadinordertosetupplansinyourlabismuchmuchlargerthanworking on microbes, single organisms and hence I think most chemists shy away from it.

Arne Skerra: At the Technical University of Munich, we have a strong Plant Science Department and I have the feeling that metabolism in plants is much better explored

than the enzymology behind it. In addition, the genomic eﬀorts in the plant area are also under-developed. So I would foresee rather fruitful research in this area, ﬁnding new enzyme functions, not only in plants but also in fungi.

Wilfred van der Donk to Jean-Louis Reymond: You showed that your artiﬁcial enzymes or enzyme analogues can mimic certain reactions that we ﬁnd in Nature. distribution is allowed. distribution

I was wondering whether you or others had looked at the chemistry that you don’t

find in Nature and that you might be able to catalyse with these structures? 9180 Jean-Louis Reymond: Our structures are not functioning as proteins but to get new only. No further No only. functions, the idea is just to figure out what you want and put it inside. We know so much about chemistry that I think you can actually design things into proteins by importing things from outside. There are all the efforts to create metallo-enzymes using metals that don’t appear in natural proteins. So we can do all of that in a

vayinstitutes.be/ completely synthetic structure like we do, or in the natural ones. As to our approach PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- to dendrimers, it reveals that you can do things very simply, so certain things are simple to plant in the structure. Except that we cannot denature these things. We are not that far yet. http://www.sol

Chris Walsh: To follow-up maybe. When I think of Nature’s dendrimer equivalent host on on host in proteins, I think of the carbohydrate chains on glycoproteins because they are the branched versions although the backbones of proteins are not. Have you made any mixed peptide oligosaccharide dendrimers to look at questions of different kind of ligand recognition? http://www.worldscientific.com/worldscibooks/10.1142/ Jean-Louis Reymond: Well we’ve played the usual game of attaching single carbohydrates at the end of the dendrimers so you get this multivalency effect. It’s very clear that you have a recognition effect that is achieved by multivalency. We can obtain this effect also in the synthetic systems. Whether the branch nature of the carbohydrate exists because of that need for multivalency, I don’t think it is that clear at this stage. I mean if you look at the complex oligosaccharides, they

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World have probably used the opportunity to branch that is oﬀered by the carbohydrates structure; Nature evolved because it could do it, because the sugar is a branching NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY structure. The dendrimers just oﬀer the branching but doesn’t answer the question of whether the branching is naturally necessary for the function.

Chris Walsh: In the spirit of dendrimers, maybe I can ask Jason Chin. Are you or your colleagues actually building dendrimer or branched proteins using multifunc- tional unnatural amino acids in site-speciﬁc ways? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 40

40 New Chemistry in the Expanding Protein Universe

Jason Chin: So, other than carbohydrates, another example that you might imagine from literature is of course ubiquitin chains where ubiquitin is attached by an isopeptide bond built from a lysine to the C-terminal of the next protein in the

chain. Ubiquitin itself has multiple lysines so you can build a number of diﬀerent

topologies. There are a number of diﬀerent lysines within ubiquitin so you can build dendrimers like structures from ubiquitin proteins and from similar and other homologues. Other people have to use amino acids that allow you to synthetically build ubiquitin just like Tom done, some of his work. . .

Chris Walsh: I meant dendrimers, not ubiquitin. distribution is allowed. distribution

Jason Chin: We are not working on building dendrimers per se.

9180 Tom Muir to Jean-Louis Reymond: I have a question about topology. We are very good at making peptides at this point and have some beautiful examples of only. No further No only. dendrimers and polycyclic compounds. Is there a topology that you would love to access using chemistry that perhaps Nature accesses, that is currently oﬀ limits? That we just don’t have the synthetic know-how to do it yet?

vayinstitutes.be/ Jean-Louis Reymond: It depends on how you deﬁne your topology. Clearly the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- example of these polycyclic natural products made from peptides, already have this polycyclic topology, and that it is possible and very simple. It’s what you learn from graphs. And then of course Nature has other thing like these lasso-peptides where http://www.sol you actually have additional dimensions of topologies with interlocked rings, where I would say supra-molecular chemistry is much more at home. That is another area host on on host of possible completely new functions that might exist when you make tied knots and things like that.

Gebhard Schertler to Jean-Louis Reymond: In your chemical space, there is a lot

http://www.worldscientific.com/worldscibooks/10.1142/ of empty space. Is it actually really principally empty or is it again evolution that has been chosen in biology or the chemistry, when we have reactions to make polymers, that actually then reinforces certain branches to be very strong? I mean the DNA branches are very strong; the peptides branch is very strong. But we saw for example when we have these cyclic peptides, then we get rigidity. If you really ask for new chemistry shouldn’t it be in this empty space?

Jean-Louis Reymond: Of course you can do this. The pretty picture I showed is of ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World course a representation of chemical space. We have this front page of the conference NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY booklet with the painting “ceci n’est pas une pipe” so this is all we do also in our representations. It all depends on how you describe it. But clearly, for example

if we map the bicyclic peptide on the map, they clearly move to the right so we actually ﬁll a completely new area of chemical space. So clearly there are a lot of opportunities in studying molecules that have diﬀerent properties by the nature of how they are made. Then you have to think carefully because I think the wonder October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 41

Session 1: Discussions 41

with proteins is for example that they have just the right balance for very high diverse functions. You might also just take the lesson from Nature by going in that direction but again, hybrid molecules are highly interesting as well and it is a

change for synthetic chemistry I would say.

Discussion among all attendees

Gebhard Schertler to Tom Muir: In GPCRs, we have phosphorylation on the C- terminus, sometimes in the loop and there was a proposal of a kind of barcode distribution is allowed. distribution for this modification. It would be established by different kinases possibly putting different modifications onto the GPCR. Now if we look at the modification field in

9180 general, is this barcode maybe a wider idea and how would we figure it out? Could we just do something which people are doing in computational sciences now, look only. No further No only. for patterns? So if we have a method to look at modifications in a large scale like we do sequencing, could we then look for patterns and could we, for example, change the signaling state and see how these patterns change? And could we actually see if there are any systematic changes in these modifications? vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Tom Muir: I think that is an interesting idea that there might be patterns. There are some analytical challenges to identifying them. Unless they are right next to each other in sequence and you can get them on a single peptide when you do mass http://www.sol spectrometry, when they are distal in a protein sequence and you are averaging over a population it is hard to know whether or not they’re all coming from one discreet host on on host polypeptide or whether they are from different members in a population. So this is certainly an analytical challenge to identify them in the first place. But assuming one could do that in some fashion, the question would be “Do they actually signify some defined output and how do we figure out what it is?” Is that essentially what

http://www.worldscientific.com/worldscibooks/10.1142/ you’re asking? If one were able to correlate those patterns with some phenotype in the cell, coming back to chromatin where we work, if it correlates with a particular type of gene transcription, we label the gene transcript. Maybe you can infer that a pattern does that, which is certainly part of what people are doing in that area. It doesn’t tell you anything about the mechanism with which that might happen of course. This again brings us back to this business of getting hold of reagents that actually have that pattern, to study the behaviour by structural methods or

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World by biochemical methods or however you may want to do it. You need to get beyond

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY correlations and I think for that you need good old-fashioned biochemistry and you need to be able to make the reagents and that’s a challenge.

Christopher Walsh to Kurt Wüthrich: Just a comment, a thought. The C-terminal tail of RNA polymerase is phosphorylated by different kinases and there is very good evidence in yeasts at least, from genetics and biochemistry, that in fact the multiple phosphorylations at specific serines dramatically affects the behaviour of the polymerase. So that might be a very good place to start analyses or calculations. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 42

42 New Chemistry in the Expanding Protein Universe

Kurt W¨uthrich: Well, I grew up in an environment where peptide synthesis was king, with Professors Robert Schwyzer and Josef Rudinger, at a time when all the Swiss pharmaceutical companies had big teams working on peptide synthesis.

ACTH, oxytocin calcitonin and growth hormone were big names, and for us it was

glucagon. Then peptide chemistry literally disappeared, all the peptide synthesis laboratories were closed in the pharmaceutical industry in Basel. Today, we are witnessing a rebirth. In the old days it used to be that the synthesis of cyclic peptides was a particular challenge, but now we see that with both biosynthetic and chemical methods, it seems to be daily routine that you make mutlicyclic peptides. I have 2 questions. One is: “How do you compare the potential of synthetic and distribution is allowed. distribution biosynthetic methods to generate these complex peptides?” And then a question

for the bioinformaticians: “How do you handle peptides, and in particular cyclic peptides?” 9180

only. No further No only. Tom Muir: When comparing the synthetic approaches and the biosynthetic ap-

proaches, obviously with the synthetic chemistry one is not constrained by the natural amino acids so you can use unnatural residues and unnatural backbones and unusual topologies that we heard about earlier. There have been great strides though in the biosynthetic, even ribosomally synthesized, cyclic peptides. It is now vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS possible to generate these using various protein chemistry tricks involving proteins

- called inteins that might work that some of you may be familiar with. You can generate quite large libraries of cyclic peptides, which are genetically encoded using

http://www.sol these sort of reagents. This obviously then leads to a selection and all the fun things that are associate. So the biosynthetic approaches have the advantage of ease of

host on on host synthesis, it’s easy to encode a library of cyclic peptides at the DNA level, but you are constrained by what you can put in. You simply wouldn’t be able to make easily some of the wonderful architectures that Jean-Louis or Wilfred, can generate using chemistry.

http://www.worldscientific.com/worldscibooks/10.1142/ Wilfred van der Donk: I think that summarizes pretty much my thoughts as well. Perhaps one thing to add is that using biosynthetic methods, I would also call out Hiroaki Suga approaches to bring in non-natural amino acids. Those are perhaps the best to get lead compounds because you can make libraries, but then synthetic chemistry can be used to optimize those much better because you have a larger structure or functional space that you can explore with synthetic chemistry. So I think maybe a combination of both is the best of both worlds. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Chris Walsh: To go back to your comments historically about the early days of peptide commercial triumphs, that was the low-hanging fruit. Those were small peptides known to be active as hormones, ACTH a little bit bigger. One can look

at the ﬁeld and say very few peptides of any size beyond 20 residues have been commercialized, it’s an interesting problem. I do sit on the board of Ironwood Pharmaceuticals, they have commercialized their 17mer for GI and of course the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 43

Session 1: Discussions 43

Roche-Trimeris group had a 37 residues anti-HIV peptide but it didn’t sell eﬀec- tively. Hard to tell whether cyclic peptides and the new generations will turn out to be a commercially interesting set of targets for the pharma industry. I would be

interested by what other people think. Maybe we can go to the bioinformatics.

Adam Godzik: The bioinformatics of peptides is way behind the bioinformatics of full proteins. Not much is known about these peptides. They could just be an overlooked part of human genome. I read recently some papers which started from environmental proteomics where numbers of peptides secreted by the human host is going up to the hundreds and some of them are not even coded in the distribution is allowed. distribution

genome. Our knowledge about this is very scattered, we know few families and

they’re characterizing hundreds of them now. So it’s a completely new area, which I think is not nearly as developed as everything. There are no databases for them, 9180 there are no good tools because we have problems with recognition of similarity

only. No further No only. between such peptides. I think it’s a huge area to be explored but not much has

been done and not much research exists yet.

Christine Orengo: I completely agree with Adam. I can’t think of any resources and it seems to me an ideal opportunity for young chemists or young chemical vayinstitutes.be/ informaticians to develop something like this. I would think if the problem was PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- more tractable in some way, than what we have to do with protein structures and sequences, I could imagine the tools beingsomewhateasiertodesigntocompare

http://www.sol these compounds. I mean there are good methods for comparing similarities.

Adam Godzik: We struggled with this a little. I was part of a project of meta- host on on host proteomics on human skin and human gut and we discovered a lot of peptides, which we can track back to being produced by humans. But we realized there was no catalogue of them and for many of them we don’t know whether they are functional peptides or just pieces of proteins, which happened to be in the neighbourhood. http://www.worldscientific.com/worldscibooks/10.1142/ Some results suggest they’re actually functional peptides. More is known about this in flies because I remember there was a big project where they went and identified peptides secreted by flies. Many of them have antimicrobial properties for instance.

Tom Muir: There is some interesting new data coming out on so-called “small open- reading frames” or SORF’s where these are not processed from larger proteins but there are the small genes that maybe have been under the radar in terms of the size.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World The analytical tools, the mass spectrometry, being used to detect these in cells are

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY being developed. A lot of these peptides do seem to be around and persist for quite some time in cells, implying that maybe they have some function but the functional aspect of what these things are doing is somewhat embryonic.

Jason Chin to Wilfred van der Donk and Chris Walsh: We thought about how to catalogue and categorize and how to understand mechanisms. We can do that endlessly but one question I have, just as a thought, is where are we with really October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 44

44 New Chemistry in the Expanding Protein Universe

understanding the level of functions? For example in the natural product area, can we recreate ecosystems and understand what the functions of these are? Is that something that is worth investing time?

Wilfred van der Donk: I think that knowing mode of action is of course something

we are always looking at not only for the reason that you pointed out but also because any type of target that you identify could be potentially an interesting target for other molecules, not necessarily your molecule that might not have great properties. It’s not so easy to identify targets and we have found that the hard way. You often get many possible targets or leads to each target. For some molecules you distribution is allowed. distribution

can look at, you can already guess what the target will be based on what we know,

but the great majority of molecules we know that are antimicrobial, how they work, we often don’t know. 9180 Chris Walsh: I just want to relate one example. There is a molecule called surfactin, only. No further No only. a lipoheptapeptide produced by Bacillus subtilis, which was identiﬁed by its ability to lower the surface tension dramatically of ﬂuids like water. So it is called surfactin and for years its mode of action was not known, it is a long chain acylheptapeptide. Two of my Harvard colleagues were studying the interaction of this particular vayinstitutes.be/ Bacillus strains with Streptomyces and they observed that the Bacillus put out a PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- peptide, which blocks the growth of Streptomyces. It turned out it was a surfactant only turned on under speciﬁc circumstances. They also realised it functions as a

http://www.sol potassium channel and so, this potassium signalling had turned on certain kinase genes. The Bacillus caused the diﬀerentiation pattern. So it’s clear that if you

host on on host want to study the behaviour of genes that are not normally turned on, that is are conditional, one should not study them in pure cultures in a laboratory. It’s still an open-research question to ﬁgure out what kinds of microenvironments you want to reproduce. That is a sort of a nice example. Surfactin was used actually to reduce the viscosity of oil in oil wells; you can pump up crude oil for decades before the http://www.worldscientific.com/worldscibooks/10.1142/ physiological functions was detected.

Richard Lerner: Chris, if you get a mixture of peptide, how do you know it is not just junk, some sort of proteolysis? I think a very good initial cut would be those peptides that are C-terminally amidated because I don’t know of any situation where Nature is going to do work to C-terminally amidate a peptide that isn’t in some sense being important. I remember in the early days of the Amylin Pharmaceutical ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World company, when during the dark days, when people wondered if amylin was real, or

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY some artifact. The share price was 25 cents. I said if it’s C-terminally amidated it must be important and I don’t know any company or any research group that has done the sort of the modern and analytical world a screen for C-terminally amidated

peptides.

Chris Walsh: To your point Richard, I think almost all of that post-translational amidation reﬂects the processing of precursor proteins. So these really are not October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 45

Session 1: Discussions 45

primary transcripts. I think Tom Muir was also referring to work for example of Jonathan Weissman. Ribosome proﬁling showed you the prediction of primary proteins which are anywhere from 40 to 100 residues which had been missed in

sequencing and whose functions are being worked out one at a time. But I can go

with, most of the C-terminally amidations come from hydroxylation and then loss of the last residue. So it’s a purposeful post-translational processing.

Richard Lerner: A lot of people have tried to make an antibody which could be a great tool to pull out interesting peptides from a mixture. It has not been successful.

Brian Roth and Don Hilvert: I’d like to expand on that. We thought a lot about distribution is allowed. distribution

deconstructing the activities of peptides and small molecules that are in the brain for

instance to ﬁnd their molecular targets and what we decided was that two things

9180 were needed. The first was basically a robust genome-wide way to interrogate the targets that is scalable. We now can basically look across entire families of only. No further No only. receptors or kinases or whatever in a very robust and relatively cheap fashion. But the problem, the stumbling block that we have is to get compounds that are likely to have biological activity and with the peptide, it’s a very difficult problem. So you can imagine if you do a mass spectrometry based approach, you might find vayinstitutes.be/ thousands of these potentially biologically active peptides that are secreted from a PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- particular neuron. And you don’t know, I mean some of them may be amidated and that may be a way to pull them out but if you want to do this from a unbiased

http://www.sol approach, basically to screen the entire home of peptides against all of the potential targets that they might interact with, you need to have some ways, cheap and robust

host on on host way of making these all in quantities that are sufficient to interrogate their biological activities. We’ve tried to do this over the years with peptide chemists that basically are mining genomic data and coming up with what they think are biologically active peptides identified from genomic and mass spec data. The stumbling block is always to get sufficient amount of material to actually test. Each one of those can be a http://www.worldscientific.com/worldscibooks/10.1142/ relatively difficult project in itself. So if there is some way of basically robustifying this so that it can be done in an unbiased way, I think we could really push things forward. The same relates to small molecules, we’re now realizing that many of the known and unknown intermediary metabolites are actually ligands for receptors or can modify enzymatic activity. Again the problem is not identifying them which can be done relatively simply with the available analytical techniques but to obtain sufficient quantities of them to actually test. This is where chemistry comes in. We ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World can imagine one of those can be a project but if you want to make thousands of NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY them, it becomes a really monumental task.

Don Hilvert: One potential problem with the bioinformatics approach to iden-

tify bioactive proteins is the extensive post-translational post-synthetic modiﬁca- tions that we have heard about. In that context, to what extent are these post- translational post-synthetic events deterministic? Does the ﬁrst event dictate the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 46

46 New Chemistry in the Expanding Protein Universe

next and so one? Obviously in the case of the histones, if you acylate a particular lysine, you can’t then methylate it, but there are other sites that may be inﬂuenced. That may also be true in smaller peptide systems as well.

Tom Muir: The idea that one modiﬁcation becomes another one is pretty well

established, at least in the chromatin area. Something that is called crosstalk and it’s clear that the enzymes that install these things require initiating modiﬁcations to work. They are either not recruited properly or not allosterically activated properly. Either way, they will not work on unmodiﬁed template. That crosstalk is a very important part of regulation in these systems and very common. Similarly for distribution is allowed. distribution

removing modiﬁcations that is just an installation taking the right ones oﬀ, there

needs to be other constellations or other modifications to do it properly at the right place. So that is an extra layer of complexity to the whole business. 9180 Wilfred van der Donk: So with the biosynthesis of these natural products, you find only. No further No only. both. We find cases where chlorination will only happen if the molecule is already cyclized and otherwise the chlorinase doesn’t touch it. We find other examples where a tailoring enzyme works on the linear peptide just as well as on the already cyclized peptide. So it seems to be sometimes required and in other cases not. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Chris Walsh: In the kinase field, there are many examples where the first phosphorylation could be stochastic. There might be 500 possible substrates in the cells but the following phosphorylation may require that first one to have been installed. http://www.sol So there are cases where it can become more deterministic as one goes to a second and a third post-translational modification step. Of course there is also the issue host on on host that it’s very rare for a site to be fully modified in a protein, that is to say it is modified in a given protein but if there are thousands of such protein molecules in the cell, there will be a gradation in terms of mole fractions phosphorylated or acylated which both complicates things and explains stochasticity for example. http://www.worldscientific.com/worldscibooks/10.1142/ Tom Muir: To add to that, there is an important role for informatics here because many of the proteins that install modifications contain domains whether they can recognize the same or different modifications and you can identify these domains from sequences. If you take a step further, obviously many proteins are part of large complexes, but then those complexes are maybe domains that recognize particular PTMs. Based on that information, you can come up with hypotheses that certain

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World patterns may be necessary for a given complex or enzyme complex to work. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Jason Chin to Wilfred van der Donk and Tom Muir: So the idea that, in Wilfred’s case, maybe doing highly parallel biosensors and, in Tom’s case, highly parallel high-

throughput biochemistry to either get a functional to make a collection of natural products. I’m curious if both of you would be able to talk a bit more about how in practice you think that is going to be achievable? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 47

Session 1: Discussions 47

Wilfred van der Donk: So there was a call by NIH this last year for a merge of synthetic biology and natural products biosynthesis. The goal was simply to scale up what we are doing now in genome mining and not to one molecule at a time but

doing it hundreds if not thousands at a time. And so that has brought together a

large number of synthetic biologists with natural product biochemists. We also put in a proposal as a team. So yes the idea would be to essentially use the merging tools of synthetic biology and expressing proteins in non-natural hosts and doing that in a robotic way. And that’s what we are applying right now to our genome mining eﬀorts. distribution is allowed. distribution

Tom Muir: Our focus has been more on in vitro biochemistry in this regard. So we

know we can make a lot of these proteins in bulk. The question is “how do you do it in parallel?” I think a key part of this is going to be integrating the synthesis 9180 with microﬂuidics, which is something my lab is doing quite a bit. It is a matter

only. No further No only. of doing it in parallel and scaling down. Engineering a proteic-ﬂuidic approach is

going to be an important part of generating all these things in parallel. Of course, you get smaller amounts of materials, so then you have to start to build at the back end of this, analytical approaches that allow you to manipulate and analyse tiny amounts of proteins, which is another part of the puzzle. But certainly in terms vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS of the manufacture of these things, ﬂuidics I think it’s an important part of the

- solution.

http://www.sol Jason Chin: Can I follow up and ask whether and to what extent there might be scope for doing your sort of work using, or not using hosts, essentially, as in vitro

host on on host experiments and how realistic that is?

Wilfred van der Donk: We do all of our mechanistic work in vitro but when it comes to genome mining, we do other things, mostly in E. coli, a little bit in yeasts, and a little bit in Streptomyces, simply because you need quantity. At the end of the day, http://www.worldscientific.com/worldscibooks/10.1142/ once you get your new natural product, you still have to determine the structure and actually we find that that might well be the biggest hurdle to scale this up, to determine the structure of the final product and to make sure that you actually have the final product because it might be just an advanced intermediate. Those are hurdle steps we would face.

Lode Wyns to Christine Orengo: We are overwhelmed by the explosion of sequences,

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World up to a point that we should maybe hold the sequencing. Now there are so many

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY proteins looking for a function, I wonder what the reciprocal exercise will do? We know about functions and there are many enzymes doing particular functions. Sta- tistically, can you do the inverse problem looking for a function and see which folds, which families are connected to it and if so, what is coming out of such an exercise?

Christine Orengo: I think it’s diﬃcult answering that. When we analyse families, and when we analyse the types of functions they are able to support, we can see that there are many more families than the functions we observed. So in other words October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 48

48 New Chemistry in the Expanding Protein Universe

the same function can be performed by multiple families in different ways, with different geometries in the active site. We could ask the question for this function “what are the various different ways in which you can engineer this function, and

what are the diﬀerent ways in which Nature has done that? I think that would be

an interesting use of the data. You’ve alluded to this explosion of sequence data. I think I disagree with the comment that is overwhelming. I think that actually it’s very easy for us to organize quite a large proportion of it. Maybe 80 or 90% of the sequences that Adam has alluded to as well are distant homologues of what we have seen already. We can very easily catalogue this new information. So I don’t think it’s overwhelming in that sense, I think it’s actually very exciting because distribution is allowed. distribution having done that, we can then start to see what are the changes that we are getting

in the active site for this particular scaffold. So we can look for positions that are hot spots, we can look for context in the active site where we can see conserved 9180 or varying electrostatic and things like that. So we can ask lots of questions very only. No further No only. easily from these data. Having more information does give us a lot more insights into ways in which you can change the chemical environment for particular scaffold. So I hope they continue with the sequencing rather than stopping it. We are not finding too much of them at the moment. vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS Judith Klinman to Christine Orengo: I wanted to comment on your search engine

- where you focused on functional groups that can be easily identiﬁed to play a role in catalysis (acids, bases, catalytic triads), whatever it is that would deﬁne

http://www.sol a particular enzyme reaction because so much more of the protein is going to be involved in catalysis. And so, going beyond this very small and constrained region

host on on host and then moving out by degree — I don’t know to what extent you’ve tried that. . . We know so much now about the impact of remote mutations on catalysis, the role of protein dynamics and other kinds of features. And I think that is also interesting in the context of what you just said about convergent evolution, where you have different scaffolds coming together to give you the same activity. What are the http://www.worldscientific.com/worldscibooks/10.1142/ roles of rates in these different families and can you use that information to guide you in terms of structural boundaries and where you look? So, it’s a question and comment at the same time.

Christine Orengo: It’s an interesting question and you’re right. It’s not always the active site that you’re interested in. And the example I showed of RuBisCo where we used all the sequence information we had in the evolutionary tree, it allowed ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World us to ﬁnd these sites under positive selection where we could see that frequently NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY there were mutations in these sites that were altering the eﬃciency of the enzyme. Very few actually are directly in the active sites, they are in a sort of next shell

around the active site ensuring that conformation or adjusting the conformation or geometry around the active site. They were also in the interface between some subunits as well, in fact aﬀecting the allostery. What was interesting was we could see multiple routes to achieving this eﬃcient enzyme. So Nature found many ways of October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 49

Session 1: Discussions 49

adjusting and tuning these positions that could have some inﬂuence on the chemistry ultimately.

Judith Klinman: Yes I think the importance of subunit interactions is something that needs more attention for sure, and how that is directed into the active site.

I did have a question for you, when you talked about the change in eﬃciency of RuBisCo, does that relate to the ability to discriminate between CO2 and O2?Is that what you mean or is it the turnover number.

Christine Orengo: Yes it is quite complex because there is this competition between oxygen and carbon dioxide for respiration and photosynthesis. And these plants that distribution is allowed. distribution

have become more eﬃcient, it’s because they had actually reduced their aﬃnity for

carbon dioxide. And by doing so, that actually increased the turnover, it’s inversely

9180 proportional to their aﬃnities. So that’s really what I meant, that was the eﬀect that we saw of these mutations. only. No further No only.

Richard Lerner to Adam Godzik: Two practical questions maybe to any of the panel. So in evolution we talk about when that disappeared and when that appeared, and so forth. Do we know when folds appeared? And is there a single case where we

vayinstitutes.be/ actually know what the order was of acquisition of folds? PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Adam Godzik: I think that folds appeared in the time we cannot really probe with tools for evolutionary analysis. They all appeared somewhere around when the ﬁrst http://www.sol cellular life probably appeared. So when we try to track evolutionary folds, it’s very hard to say anything about it because it is really too far for us to say anything. host on on host They are some folds, which appeared later but I think there are very few. When we ask exactly this question, I think we found like 5 or 6 folds, which are speciﬁc to eukaryotes.

http://www.worldscientific.com/worldscibooks/10.1142/ Richard Lerner: But you know proteins that used to be something, pick up one more fold and become something else?

Adam Godzik: But we see. . . for instance especially multi-domains proteins acquiring certain folds so we go track back to the core enzyme.

Richard Lerner: No but very speciﬁcally, is there an example of a protein that picked up a new fold and became something else? One fold? ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Christine Orengo: Perhaps I could just add one thing to what Adam said. I think if you use the phylogenetic trees, if you trace back, you do see that you can identify perhaps what you would consider to be the most ancient folds and they’re often the most highly populated folds. There are lots of superfamilies that adopt these kinds of structure scaﬀolds. If you look at them, they make sense in some ways in that they are highly regular structure arrangements with quite a lot of hydrogen October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 50

50 New Chemistry in the Expanding Protein Universe

bonds. Here for example, the β-barrels or the α/β-barrels. You can understand why these structure arrangements have great plasticity to sequence change, so they can support wide range of sequences. In fact you can theoretically predict that these

types of structure arrangements would be able to support much more sequence

divergence than some of the more irregular structure arrangements. People who have looked at the emergence of new folds using the phylogenetic trees, have shown that, to a certain extent, these new folds are less ordered. They have a more irregular arrangement in secondary structures. More recent ones also seem to be more unique to species, so they have not been so widely used across the tree of life. I don’t think there is much more information on that at the moment. The problem distribution is allowed. distribution is that the structural data is still very sparse compared to the sequence data, which

makes it a bit diﬃcult to answer these questions using phylogenetic information, because you can’t build very good phylogenetic trees just with a structural data. 9180

only. No further No only. Gebhard Schertler to Christine Orengo: Can I make just a little comment to this?

So you are looking at sequences but if you look at folds, kind of three-dimensional context become very important. There is a lot of small amino acids (glycine, lysine, leucine), they pack and if something is hydrophobic, they look conserved. But somehow, they are not fully conserved so this is diﬃcult to analyse. In GPCRs, vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS we have used a kind of network analysis to actually get on this, it’s in a Nature

- review recently. And what we can do with this, it’s actually we can kind of remove us a little bit from the sequence but work with proximity. And I think if you would

http://www.sol look at fold evolution, so how do you transform from one fold to the other, you would need the kind of tool like that and the sequence might not be the ideal thing

host on on host because many things are held together by this hydrophobic core of the proteins which is made out of these small amino acids, which are really not so distinctive in sequence analyses. So when you reduce the protein to kind of contact and abstract it to this, you are actually looking at this space in a diﬀerent way and you don’t

http://www.worldscientific.com/worldscibooks/10.1142/ need every structure, you can use your structure and the sequence variability and use that in the network analysis. And I actually hope that brings more structural insights in the comparison. That’s why we started doing that as an idea but I don’t think we did it very well.

Christine Orengo: If I could briefly respond to that. I think you are right and you are sort of translating your sequence code into a set of properties that you put then against structure scaffold. There are other people in the room involved in protein ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World design and I know there was a competition at one point and I don’t know if it NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY still runs to try to convert one structure into a completely different structure which uses the idea of an alphabet of residue properties and switching those properties

dramatically to try to engineer a completely diﬀerent fold. There might be people in the room who know about those competitions? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 51

Session 1: Discussions 51

Bernard Henrissat, Adam Godzik and Christine Orengo: I would like to say that I agree very much with what Christine Orengo said about the fact that the number of sequences is not so much of a problem. There are bioinformatics methods that

are really designed to handle them. Where I have a problem is with the storage of

the knowledge, of functional knowledge. There is no way that I know where this is done in a systematic manner to capture this information and to store it just by analogy to what is done with three-dimensional structures. There is a PDB with some rules that we can follow easily and we must follow if we want to publish. And for functional data, it’s not so good, it’s not good at all. And we have some systems that have changed of purpose. I suggest the EC numbers that were not designed distribution is allowed. distribution for bioinformatics and they are being used now for bioinformatics. And these EC

numbers however imperfect they are, they’re only for enzymes and proteins are not only enzymes. We are missing also descriptors for binding relationships between a 9180 protein and a ligand. only. No further No only.

Adam Godzik: Overall you’re right but it’s not because of lack of trying. So there are eﬀorts such as gene ontology, which tries to map literature information into formalized function description. I think EC system was also an inspiration for the gene ontology. There are several groups working seriously but I agree that overall, vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS as compared to the amount of structural information captured in PDB, the amount

- of functional information available in literature captured in this classiﬁcation is relatively small. But there is also a problem with distribution of it. So I agree with

http://www.sol you, we find routinely papers from ten years ago, which defined the functions of proteins that were not captured anywhere. So there are efforts to do this, but I

host on on host agree that this is way behind our capturing structure information.

Christine Orengo: I think a lot of money has been put in to the gene ontology classiﬁcation. They are trying to expand it even further to reach out to communities working on particular pathways of chemistry and get them to actually annotate the http://www.worldscientific.com/worldscibooks/10.1142/ proteins within this resource. But obviously, that is quite a big initiative and it’s quite slow, but another quite signiﬁcant problem is with querying the data as well, it’s much harder to query than the EC information. So that’s another problem, it’s not just organizing it, but organizing in the way that you can query easily is another challenge.

Ian Wilson to Adam Godzik and Chris Walsh: So given that there appears to be a ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World lot less folds than initially anticipated. Adam, you talked about 1300 we currently

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY know but I think you backed oﬀ on that in a previous question saying there is probably not that many more folds to ﬁnd than we already know. So when is it that we’re going to be able predict from the sequence alone, from any sequence,

what fold a structure is going to adopt from the sequence?

Adam Godzik: First let me clarify that there are diﬀerences. So there are 1300 or so, plus/minus a hundred folds we have found. There are several thousands, which October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 52

52 New Chemistry in the Expanding Protein Universe

are mathematically possible, and to rephrase what David Baker did, they designed a sequence to fold into a new fold, which is not seen in Nature. They synthesized it and they did. So for many of these folds, which are technically or theoretically

possible, there are no physiochemical reasons for them not to exist. Nature simply

hasn’t found it. And this is one of the puzzles here, why there are so many millions of sequences and there are still possibilities of how they can shape and they were not used. One of the possibilities here is that Nature simply was happy with what it found. The second part of the question is again, there are big eﬀorts to go there and David Baker’s group and few other competing groups have really made big progress. So at this point, there are sequences for which we can predict structure. distribution is allowed. distribution When it could be possible for all the sequences, . . .

Chris Walsh: about 90%. . . 9180 Christine Orengo: I think that’s possible. . . only. No further No only.

Adam Godzik: Basically the problem is that folds are very often deﬁned by a delicate balance between diﬀerent competing forces and getting this balance exactly right requires to both contribution to be calculated very accurately. So as I’ve just

vayinstitutes.be/ quoted David Baker, he can predict easily exaggerated folds — so folds, which PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- are dominated by one type of force. This can be easily predicted. But especially proteins that are really on the bridge between two opposing forces are very diﬃcult to get correct. http://www.sol

Richard Lerner: But isn’t it a problem that you’re looking at the thermodynamic host on on host outcome and you have no idea whether any fold would be kinetically possible in a context of the thermodynamically ﬁnished protein?

Christine Orengo: I think that also brings to another issue with the kinetics and

http://www.worldscientific.com/worldscibooks/10.1142/ why we maybe don’t see so many of the folds that we predict should be stable and accessible to Nature. And some of the ones that we see very very frequently when you analyse their topologies, you can understand perhaps why they would fold very rapidly. They’ve got a lot of local contacts between secondary structures and things like that. So I think kinetics is another important consideration in perhaps explaining why we have this limited fold set.

Rudolf Glockshuber to Wilfred van der Donk: I would like to come back to the ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World bioactive peptides and I also have a question to the bioinformaticians. So we heard NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY that many biosynthesis pathways of these peptides are known nowadays. The enzymes introduce unusual modiﬁcations unknowingly. So do you think that there

is a potential in the long term for bioinformatics to predict the covalent structures of all bioactive peptides that an organism produces from only knowing its genome sequence? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 53

Session 1: Discussions 53

Wilfred van der Donk: I’m not a bioinformatician but currently, certainly not. An enzyme that hydroxylates an amino acid, no idea which amino acid, at which carbon, at which position and so whereas we can say from the enzymes, ok these are

probably the post-translational modiﬁcations that are going to take place. Where

they will take place? And when you have cyclic, polycyclic compounds what their topology would be? I cannot predict it right now but maybe if we have enough sequences and enough examples of solved structures, and that, I think, is going to be required as well, not of the enzymes but of the natural products. Maybe at some point we will get close but currently I would say no. But again, I am not in bioinformatician. distribution is allowed. distribution

Adam Godzik: I don’t know but I think we are looking at a sort of exponential, an explosion of possible permutations. So you can list all the possible mutations and 9180 you can calculate how many diﬀerent possibilities there are, which of them will be

only. No further No only. realised. It’s much more diﬃcult to understand and you have to know how they are

regulated and this is something which we don’t know at this point.

Don Hilvert to Christine Orengo: I was really fascinated by the example of the carboxylases that you mentioned, that have the same scaﬀold, catalyse the same vayinstitutes.be/ reaction but use a completely diﬀerent functionality. With those, was the activity PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- identiﬁed before the sequence analysis or did it emerge from the sequence analysis? Secondly where did this rare case arise evolutionarily? Can one trace its lineage?

http://www.sol Did it come from a completely different activity and then regain the carboxylase activity? host on on host Christine Orengo: So the function information came from the literature. So we’ve combined all of our family information with experimentally annotated functions that we have from either the EC or from the gene ontology. So that’s were the function information came from. And then it’s a difficult question as regards to the http://www.worldscientific.com/worldscibooks/10.1142/ phylogeny because some of the families are so diverse and we don’t have enough of the intermediates sequences to be able to trace exactly what the steps are. It is a difficult question as to whether it is convergent evolution. These arrangements have a reason and have been efficient, have been preserved within an organism. Whether you have a case of a where you have mutational drifts and it lost some of its chemical activity, it reduced efficiency but then that is subsequently recovered by further mutations in different sites in the pocket. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Don Hilvert: It went through an intermediate with completely diﬀerent activity.

Christine Orengo: Exactly! So I mean that is why for me I would like many more

sequences because then it’s easier for us to trace what the steps might be in these diﬀerent routes to diﬀerent answers or how you can facilitate this chemistry given this sort of structure arrangement. But we actually do need more sequences and actually, more importantly we need more structural data as well. That’s why the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 54

54 New Chemistry in the Expanding Protein Universe

structure initiative in genomics in the States was so exciting because they were allowing us to answer some of these questions and try to solve structures with some of the intermediates sequences.

Jason Chin to Chris Walsh: So Chris, I think one of the things that you mentioned

in your talk was the radical SAM superfamily having on the order to the 104 open- reading frames and this might be a good place to look for new activity or new chemistry. I wondered with all of your experience whether there is chemistry that you imagine might be in biology that you haven’t seen experimentally and things like transformations that you’re particularly interested in looking for? In that family distribution is allowed. distribution

or any other family more probably.

Chris Walsh: I believe oxidation chemistry is amongst the most interesting that

9180 happens biotically or abiotically. I think we’ll here from Frances Arnold about some P450-mediated chemistry that is novel. And I guess I would just go back and only. No further No only. say I think most of the sort of nucleophilic-electrophilic reaction manifolds, that happen in Nature are well understood and predictable. The electron radical carbon chemistry I think it has been very slow to come up in part because that happens in microbes that are anaerobic under simple biological niches and because the enzymes vayinstitutes.be/ have been oxygen-sensitive and so not amenable to study under aerobic conditions. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- So I put my money on oxidative transformations that are yet to be worked out but I’m sure Frances and others may, during the course of this afternoon come back to

http://www.sol that question.

Tom Muir to Chris Walsh: What about the fluorinationfor example? So everybody host on on host has a little chemist favourite modifications at the moment right, or maybe forever. To my knowledge there is only one fluorinase known right, the SAM-dependent fluorinating enzymes. Maybe that’s out of date but that they are not so many, right? Is that surprising? http://www.worldscientific.com/worldscibooks/10.1142/ Chris Walsh: The comment is about fluorination and let me put it in a category of halogenation more generally. In fluoride, it’s the most difficult of the halogens to incorporate into organic scaffolds, in part because it’s so nucleophilic and you have solvation energy problem to get it de-solvated. So yes, as far as I know, there is only one class of fluorination that has been discovered. By contrast, there are 4 different chlorination families that have evolved and have used different kind of

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World cofactors. So I think it’s the diﬃculty of de-solvating ﬂuoride ion that is the problem

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY there. There are brominations very common in marine organisms because of the amount of bromide in sea water and I think there are only about a dozen iodo compounds, iodotyrosine and iodothymidine of course being the most famous that are found. So there are some interesting niches there perhaps in halogenation. But last comment, chlorination most of it is oxidative and you cannot oxidize ﬂuoride ion under biological conditions.

Don Hilvert: It might be worth mentioning Michelle Chang’s recent work incorpo- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 55

Session 1: Discussions 55

rating ﬂuorine using the non-ribosomal peptide synthesis or polyketide synthesis, which I think is really beautiful. Clearly, it needs a lot of work but I think it’s a very very nice study and shows where we can go in this area.

Chris Walsh: Other comments?

Judith Klinman: Just this idea of aerobes and anaerobes, radical SAM. Someone mentioned the possibility of doing bioinformatics in the division between the anaerobes and aerobes. There are radical SAM enzymes that appear in aerobes and no one really mentioned that and how they succeed because these are so rare and intolerant, whether there is any information through bioinformatics that could give distribution is allowed. distribution

us a clue. It’s kind of a crossover between the anaerobic and the aerobic world. I

just put that out as a curiosity, I don’t really have any insight to say beyond there

9180 but it is a puzzle and something to just put out. only. No further No only. Chris Walsh: There are many microaerophilic bacteria that practice anaerobic chemistry so to speak. All the non-heme mononuclear iron enzymes when isolated are very oxygen-sensitive. Some of them are turning over only 2 to the 5 times before they kill themselves. And yet presumably in microaerophilic organisms, they

vayinstitutes.be/ run thousands of turnovers. So it’s a compartmentalisation and control of oxygen PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- tension in the organism maybe.

Aled Edwards: I ﬁnally give some perspective just on the protein universe. I think http://www.sol the ﬁrst meeting we held on it was probably 1998 when we were thinking about the PSI at the very beginning as we’ve tried to classify. Then the DOE had a meeting host on on host I went to in 2002 about the unannotated proteins in every genome that they were sequencing and the message was we need more and more experiments but it’s too expensive. In 2006, I went to another meeting talking about the unannotated proteins every new genome has — the proteins we don’t know what they do, we

http://www.worldscientific.com/worldscibooks/10.1142/ should do something but it’s too expensive. And in 2010, the same, the PSI went for new folds, it was fantastic but too expensive, stop doing that and do some real biology. And now in 2013, we are all talking about the same problem, about the same fraction of the genome. So I think what we can do actually is to ﬁgure out how to get it done instead of talking about these unannotated ones.

Chris Walsh: John Gerlt, you better ﬁgure that out in the next hour and a half. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

July 25, 2013 17:28 WSPC - Proceedings Trim Size: 9.75in x 6.5in icmp12-master

This page intentionally left blank distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 57

Session 2 Exploring Enzyme Families and Enzyme Catalysis distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 58

distribution is allowed. distribution

Illustration of the iron-heme center of a cytochrome P450 enzyme

9180 engineered by the Arnold laboratory at California Institute of Technology to catalyze non-natural reactions. only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 59

MECHANISTIC ENZYMOLOGY AND CATALYST DESIGN

DONALD HILVERT Laboratory of Organic Chemistry, ETH Z¨urich CH-8093 Z¨urich, Switzerland distribution is allowed. distribution My view of the present state of research on exploring enzyme

families and enzyme catalysis 9180 Enzymology is a mature ﬁeld of study. Over decades, thousands of enzymes have only. No further No only.

been catalogued and characterized; thanks to intrepid biochemists and microbiologists and recent progress in genome mining, novel activities continue to emerge apace. Sustained interest in this area reﬂects the importance of enzymes as the catalytic engines of the cell and as targets for therapeutic intervention. Technologi-

vayinstitutes.be/ cal advances have also made biocatalysts available for novel applications inside and PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

outside of the cell, paving the way for synthetic biology [1] and environmentally - friendly synthetic chemistry [2]. The basic workings of many enzymes are now known. Detailed structural

http://www.sol and mechanistic studies have elucidated the sequence of bond-making and bond- breaking events, the identity of key reaction intermediates, and the role of critical

host on on host functional groups. Nevertheless, explaining the origins of catalytic proficiency in quantitative terms has proved difficult [3]. A full accounting of enzyme efficiency is problematic owing to the complexity of these systems and the non-additivity of the contributing factors. As a consequence, fundamental concepts such as the roles of

http://www.worldscientific.com/worldscibooks/10.1142/ electrostatics [4] and dynamics [5–7] or the relative importance of transition state versus ground state eﬀects [8–10] are still controversially discussed. Given our incomplete understanding of protein structure-function relationships, expanding the universe of enzymes by design constitutes a formidable challenge. Not only is the number of theoretically possible protein sequences astronomically large, functional sequences are also rare. First steps toward fully automated de novo design of artiﬁcial enzymes are being taken [11], but adapting natural protein scaf-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World folds for new tasks is usually simpler than starting from scratch. Diverse strategies exist for tailoring enzyme properties [12]. Both targeted and random mutagenesis NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY can be employed, for example, to increase stability, alter speciﬁcity or create completely new function. Once identiﬁed, low levels of activity are reliably optimizable

by directed evolution [13]. Chemical guidance, provided either by structural data or computational design, can greatly enhance the chances of success of such endeavors by increasing the frequency of variants possessing interesting catalytic activities in the sample population [14]. Nascent enzymes for several abiological reactions have been created in this way [15, 16]. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 60

60 D. Hilvert

Design and evolutionary optimization of artiﬁcial enzymes has the potential to bring our understanding of basic enzyme chemistry into sharper focus. The general principles learned are likely to be transferable to increasingly complex systems and

thus support wide-ranging applications in research, industry and medicine.

My recent research contributions to exploring enzyme families and enzyme catalysis My group utilizes chemical synthesis, immunology, molecular biology, and genetics to investigate enzyme catalysis. The functional modiﬁcation of existing active sites, epitomized by conversion of a PLP-dependent racemase into an aldolase through distribution is allowed. distribution

a single active site mutation [17], is one focus. Enzymes that catalyze pericyclic

reactions, including chorismate mutase and various pyruvate lyases, are another.

9180 Exploitation of macrophomate synthase, a putative Diels-Alderase, as a promiscuous aldolase for the stereoselective synthesis of a wide range of 3-deoxyketo sugars only. No further No only. highlights the practical opportunities made possible by such research [18]. Laboratory evolution — iterative rounds of mutagenesis and screening or selection — has become a powerful tool for studying and modifying proteins. We have exploited this approach to probe catalytic mechanism, investigate the determinants vayinstitutes.be/ of protein structure, redesign protein topology, and tailor supramolecular structures PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- (Fig. 1) [14]. The evolution of a highly active monomeric chorismate mutase with all the features of a molten globule — a dynamic ensemble of poorly packed and rapidly interconverting conformers — illustrates the surprises that can emerge from http://www.sol such studies [19, 20]. That laboratory evolution is not limited to catalysis is ex-

host on on host empliﬁed by protein capsids engineered to encapsulate and sequester toxic cargo [21]. We are further extending this technology to customize protein containers into nanoscale reaction chambers and delivery vehicles.

http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Fig. 1. Evolved catalysts and constructs. (a) A binary-patterned chorismate mutase constructed from a limited library of 9 amino acids [22]. (b) A molten globular enzyme [19, 20]. (c) Encapsu- lation of toxic HIV protease in an engineered protein container [21].

Reactive functional groups, metal ions and organic cofactors can be employed to expand the capabilities of proteins [12]. For example, novel spectroscopic and redox properties of the element selenium render it a valuable mechanistic probe October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 61

Mechanistic Enzymology and Catalyst Design 61

and catalytic group. In biological systems, selenium is incorporated into proteins as selenocysteine, sometimes termed the twenty-ﬁrst proteinogenic amino acid. In the lab, we have engineered artiﬁcial selenoenzymes by chemical, semi-synthetic, and

recombinant methods. In cytochrome P450cam, for instance, recombinant replace-

ment of the proximal heme ligand by selenocysteine afforded a unique mechanistic and spectroscopic handle on reactive intermediates (Fig. 2(a)) [23]. To access an even wider range of chemical activities, coupling the unique reactivity of transition metals with the selectivity conferred by specific protein scaffolds represents an attractive strategy [24]. In one case, we attached a Grubbs–Hoveyda type catalyst to a protein framework to yield an artificial metalloenzyme that promotes ring-closing distribution is allowed. distribution metathesis [25]. As an extension of efforts to create abiological enzyme mimics,

we are also developing catalysts based on other sequence-deﬁned polymers, such as β-peptides and mixed α/β-peptides, which can fold into well-deﬁned secondary and 9180 tertiary structures [26]. Such molecules are useful for testing the tenets of molecular only. No further No only. recognition and catalysis.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host

Fig. 2. Artiﬁcial enzymes. (a) A variant of cytochrome P450cam in which the axial cysteine ligand was replaced by selenocysteine [23]. (b) A computationally designed and experimentally optimized Kemp eliminase showing unusually high eﬃciency [30]. http://www.worldscientific.com/worldscibooks/10.1142/

Exploiting the promise of computation, our group collaborates with David Baker (University of Washington, Seattle), Ken Houk (UCLA) and Steve Mayo (Caltech) on the design of enzymes for abiological reactions. Computational design is conceptually similar to the creation of catalytic antibodies [27, 28], but because it can explore protein sequence space more thoroughly, it has the potential to go far be-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World yond. This approach has already yielded catalysts for simple hydrolytic and proton transfer reactions, a bimolecular Diels-Alder cycloaddition, and multistep retro- NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY aldol reaction reactions [15, 16]. Although the starting designs typically exhibit modest activities, they are readily improved by directed evolution. Optimization

of the originally programmed mechanism is the norm, but the unexpected does occur, as evidenced by extensive active site remodeling and a catalytic residue switch observed for a model retro-aldolase [29]. Evolution of a computationally designed enzyme that catalyzes the Kemp elimination, a simple proton transfer reaction, October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 62

62 D. Hilvert

−1 −1 −1 with a kcat of 700 s and a kcat/KM of 230,000 M s shows that very high rates can be achieved in favorable cases [30]; the 6 × 108 fold rate acceleration exhibited by this catalyst approaches the exceptional eﬃciency of highly optimized natural enzymes like triosephosphate isomerase. Structural studies are important to chron-

icle how shape complementarity and precisely placed catalytic groups give rise to such eﬀects (Fig. 2(b)).

Outlook to future developments of research on exploring enzyme families and enzyme catalysis

distribution is allowed. distribution As with many contemporary scientiﬁc challenges, protein chemistry is amenable

to integrated, interdisciplinary approaches. An array of chemical, biological and

biophysical tools is opening a multiplicity of pathways to a better-nuanced under- 9180 standing of protein structure-function relationships, enzyme function and evolution,

only. No further No only. and the engineering of entirely new properties.

De novo design of enzymes for any conceivable reaction remains a distant prospect, yet there is reason for optimism that the protein universe will continue to expand in exciting new directions. Artiﬁcial enzymes are attractive alternatives to natural enzymes for elucidating basic principles of catalysis and as practical cat- vayinstitutes.be/ alysts in their own right. Bottom-up production of these agents, as opposed to PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- top-down dissection of naturally evolved systems, promises to enhance our understanding of the origins of catalytic eﬃciency. Emerging technologies, particularly

http://www.sol the combination of computation and laboratory evolution, can be expected to facilitate exploration of varied catalytic motifs and starting scaﬀolds with respect to

host on on host activity and specificity, on the one hand, and further diversification on the other. Beyond protein catalysts for relatively simple chemical transformations await more complex assemblages, including non-biological polymers and molecular machines. The creation of orthogonal genetic materials, complex signal transduction systems, biosynthetic pathways for unnatural products, artificial organelles and even http://www.worldscientific.com/worldscibooks/10.1142/ whole cells represent exciting prospects for the future. Modern challenges in human health, energy, and sustainability will certainly benefit from progress toward these aims.

Acknowledgments The author is indebted to the ETH Z¨urich, the Swiss National Science Foundation ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World and DARPA for generous support. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

References

1. P. E. Purnick, R. Weiss, Nat. Rev. Mol. Cell Biol. 10, 410 (2009). 2. U. T. Bornscheuer, G. W. Huisman, R. J. Kazlauskas, S. Lutz, J. C. Moore et al., Nature 485, 185 (2012). 3. D. Herschlag, A. Natarajan, Biochemistry 52, 2050 (2013). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 63

Mechanistic Enzymology and Catalyst Design 63

4. A. Warshel, P. K. Sharma, M. Kato, Y. Xiang, H. Liu et al., Chem. Rev. 106, 3210 (2006). 5. Z. D. Nagel, J. P. Klinman, Nat. Chem. Biol. 5, 543 (2009).

6. V. L. Schramm, Annu. Rev. Biochem. 80, 703 (2011).

7. G. G. Hammes, S. J. Benkovic, S. Hammes-Schiﬀer, Biochemistry 50, 10422 (2011). 8. T. C. Bruice, S. J. Benkovic, Biochemistry 39, 6267 (2000). 9. M. Strajbl,ˇ A. Shurki, M. Kato, A. Warshel, J. Amer. Chem. Soc. 125, 10228 (2003). 10. M. Garcia-Viloca, J. Gao, M. Karplus, D. G. Truhlar, Science 303, 186 (2004). distribution is allowed. distribution 11. B. Kuhlman, G. Dantas, G. C. Ireton, G. Varani, B. L. Stoddard et al., Science

302, 1364 (2003). 12. M. D. Toscano, K. J. Woycechowsky, D. Hilvert, Angew. Chem. Int. Ed. 46, 9180 3212 (2007). only. No further No only. 13. F. H. Arnold, Acc. Chem. Res. 31, 125 (1998). 14. C. Jäckel, P. Kast, D. Hilvert, Annu. Rev. Biophys. 37, 153 (2008). 15. G. Kiss, N. Ç elebi-Ol¸¨ cüm,R.Moretti,D.Baker,K.N.Houk,Angew. Chem. Int. Ed. 52, 5700 (2013).

vayinstitutes.be/ 16. D. Hilvert, Annu. Rev. Biochem. 82, 447 (2013). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 17. F. P. Seebeck, D. Hilvert, J. Amer. Chem. Soc. 125, 10158 (2003). 18. D. G. Gillingham, P. Stallforth, A. Adibekian, P. H. Seeberger, D. Hilvert, Nat. Chem. 2, 102 (2010). http://www.sol 19. K. Vamvaca, B. Vögeli, P. Kast, K. Pervushin, D. Hilvert, Proc. Natl. Acad. Sci. USA 101, 12860 (2004). host on on host 20. K. Pervushin, K. Vamvaca, B. Vogeli, D. Hilvert, Nat. Struct. Mol. Biol. 14, 1202 (2007). 21. B. Wörsdörfer, K. J. Woycechowsky, D. Hilvert, Science 331, 589 (2011). 22. K. U. Walter, K. Vamvaca, D. Hilvert, J. Biol. Chem. 280, 37742 (2005).

http://www.worldscientific.com/worldscibooks/10.1142/ 23. C. Aldag, I. A. Gromov, I. Garcia-Rubio, K. von Koenig, I. Schlichting et al., Proc. Natl. Acad. Sci. USA 106, 5481 (2009). 24. T. Heinisch, T. R. Ward, Curr. Opin. Chem. Biol. 14, 184 (2010). 25. C. Mayer, D. G. Gillingham, T. R. Ward, D. Hilvert, Chem. Commun. 47, 12068 (2011). 26. M. M. M¨uller,M.A.Windsor,W.C.Pomerantz,S.H.Gellman,D.Hilvert, Angew. Chem. Int. Ed. 48, 922 (2009).

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 27. R. A. Lerner, S. J. Benkovic, P. G. Schultz, Science 252, 659 (1991).

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 28. D. Hilvert, Annu. Rev. Biochem. 69, 751 (2000). 29. L. Giger, S. Caner, R. Obexer, P. Kast, D. Baker et al., Nat. Chem. Biol. 9, 494 (2013). 30. R.Blomberg,H.Kries,D.M.Pinkas,P.R.E.Mittl,M.G.Gr¨utter et al., Nature 503, 418 (2013). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 64

LOOKING IN NEW DIRECTIONS FOR THE ORIGINS OF ENZYMATIC RATE ACCELERATIONS

JUDITH P. KLINMAN Department of Chemistry, Department of Molecular and Cell Biology, and California Institute for Quantitative Biosciences, University of California Berkeley, CA 94720-3220, USA distribution is allowed. distribution

9180 Historical overview of enzyme catalysis only. No further No only. As the most powerful catalysts in Nature, enzymes are capable of accelerating rates of reaction by up to 1025-fold [1]. A satisfactory explanation for the origins of these enormous rate accelerations remains a major challenge. Structural studies provide a robust context for our visualization of enzyme active sites. It is immediately clear

vayinstitutes.be/ that the “molecularity” of enzyme reactions has no counterpart in small molecule PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- chemistry, with a multitude of side chains positioned close to the substrate-binding pocket to provide nucleophilic, acid/base and electrostatic catalysis. When this is combined with the action of active site bound organic cofactors and metal ions, the http://www.sol uniqueness and catalytic power of an evolved enzyme active site becomes apparent. Historically, one of the enduring treatments of enzyme rate acceleration has been host on on host in the context of transition state theory, with catalysis emerging, by deﬁnition, as a result of a reduced free energy barrier height. This had led to an almost exclusive focus on “enhanced transition state binding” as the physical origin of enzyme catalysis. Many investigators have pondered how subtle changes in reactant

http://www.worldscientific.com/worldscibooks/10.1142/ bond lengths and angles that occur in the transition state can produce a binding enhancement by up to 1025-fold. Additionally, the neglect of protein motions in earlier theories now seems untenable. Increasingly, investigators have been arriving at the view that the total protein structure will play a role in the tuning of active site structure and reactivity. Motions of the protein backbone and side chains have long been recognized as important for the entry and egress of substrate and product from an enzyme active site, together with the movement of protein domains and loop

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World closures that bring key active site residues into proximity with bound substrate(s).

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY More recently, another role for protein motion has emerged as an important element in catalysis. This is proposed to involve a stochastic sampling among multiple ground state structures to obtain the protein conﬁgurations that are optimal for catalysis (e.g.,[2]). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 65

Looking in New Directions for the Origins of Enzymatic Rate Accelerations 65

Evidence for active site compaction in enzyme catalysis The present discussion will be focused on two widespread classes of biologically important reactions involving hydrogen and methyl transfer, respectively. While the

details of each study diﬀer greatly, a similar picture is emerging in which barrier compaction, controlled in part by the size of strategically placed amino acid side chains, exerts a dramatic impact on the rate of the enzymatic chemical transformation step(s). C–H Activation and Hydrogen Tunneling. Studies on enzyme-catalyzed C–H activation reactions have indicated the importance of through-the-barrier, hydrogen tunneling behavior (cf. [3]). The ability to stably incorporate isotopes of hydrogen distribution is allowed. distribution at the reactive bond and to measure kinetic isotope eﬀects (KIEs) has played a key role in characterizing these systems. A number of distinctive features have emerged

9180 in relation to similar measurements for small molecules in solution [4]. First, almost uniformly for native enzymes operating under optimal conditions, the temperature only. No further No only. dependence of the KIEs is found to be very small, often undetectable, Fig. 1(A, top). Second, deviations away from optimal conditions (e.g., by mutagenesis or alteration in reaction conditions) are found to increase the barrier for reaction of the heavy isotopes of hydrogen much more than for protium itself, producing the pattern of vayinstitutes.be/ temperature-dependent KIEs, Fig. 1(A, bottom). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

(A) (B)

http://www.sol

host on on host

http://www.worldscientific.com/worldscibooks/10.1142/

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Fig. 1. (A) Depiction of the two extreme temperature dependencies of KIEs. Top is the pattern

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY characteristic of native proteins, and bottom represents the behavior under non-optimal catalytic conditions. (B) The multi-dimensional nature of H-tunneling depends on the adjustment of ground state electrostatics (a), wave function overlap (b), and donor-acceptor distance samplings (c), [4].

The behavior in Fig. 1(A) has now been observed for a large number of enzyme reactions that diﬀer with regard to substrate structure, active site cofactor and the nature of the hydrogen transfer step (hydrogen atom vs. hydride ion or proton). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 66

66 J. P. Klinman

Experimentally, mutagenesis of hydrophobic side chains has proven very informative, allowing a systematic analysis of the impact of the creation of protein packing defects on the generation of temperature-dependent KIEs.

Many theoretical models have been advanced in an eﬀort to explain the prop-

erties of enzymatic C–H activation reactions. A physical model, capable of representing the behavior of Fig. 1(A), is illustrated in Fig. 1(B) [4]. The movement of hydrogen from donor to acceptor is treated fully quantum mechanically, with a mass dependence for the rate that comes from a differential efficiency of wave function overlap, H>D>T, Fig. 1(Bb). The extent of this mass dependence is linked to the distance over which the particle moves, with longer donor-acceptor distances distribution is allowed. distribution (DADs) favoring the more diffuse wave property of H (giving rise to larger KIEs);

shorter distances diminish the gap in reactivity among isotopes, producing smaller KIEs. Of considerable importance for interpreting the experimentally observed 9180 trends in KIEs, wave function overlap is independent of temperature. only. No further No only. The temperature dependence of the rate comes solely from heavy atom motions that are required to tune the active site environment for successful H-transfer, and these represent the activation barrier. A key coordinate describes the thermal activation necessary to achieve transient degeneracy of vibrational modes in the re-

vayinstitutes.be/ actant and product wells (essential for all reactions where the equilibrium constant PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- for inter-conversion of substrate and product is diﬀerent from unity). This term is analogous to the Marcus description for electron tunneling [5], and depends on two primary parameters: λ, the environmental reorganization and ΔGo, the reaction http://www.sol driving force, Fig. 1(Ba). Separation of wave function overlap, Fig. 1(Bb), from the motions of the heavy atom environment, Fig. 1(Ba), results in a near absence of host on on host cross-terms for mass and temperature-dependent phenomena as required from the observations of temperature-independent KIEs. One feature that is omitted from the above description is the magnitude of the initial DAD. In virtually all theoretical treatments of hydrogen tunneling, the

http://www.worldscientific.com/worldscibooks/10.1142/ distance between donor and acceptor atoms must approach ca. 2.7 A˚ for protium tunneling and even less than this value for the movement of the heavier isotopes D and T. The reactive internuclear distance is, thus, more than 0.5 A˚ reduced from an estimated van der Waals distance of ca. 3.2 A.˚ An additional class of thermally- activated protein is generally invoked to create short, tunneling-appropriate DADs. These motions are referred to as conformational sampling and though introduced here in the context of H-tunneling, are likely an essential component of all enzyme

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World reactions [2].

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY SLO-1 oﬀers a classical example of enzymatic tunneling behavior, illustrating how mutation at a single conserved hydrophobic side leads to a progression from 553 Ea(D) − Ea(H) close to zero for wild-type enzyme to 5.3 kcal/mol for Ile Gly [6]. The incorporation of this behavior into the physical model for enzymatic H- tunneling is achieved via the introduction of a third coordinate to the tunneling process. As shown in Fig. 1(Bc), this involves a DAD sampling term that becomes October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 67

Looking in New Directions for the Origins of Enzymatic Rate Accelerations 67

a necessity when an enzyme loses its capacity to achieve the compact active site conﬁgurations of native protein. In contrast to the behavior of Figs. 1(Ba) and 1(Bb), Fig. 1(Bc) represents a thermally active motion that is signiﬁcantly depen-

dent on the mass of the transferred particle. The latter follows from the need to

reach shorter DADs for tunneling of the isotopically heavy particles: once the controlled tuning of active site geometries within native enzymes has been disrupted, a behavior more characteristic of solution reactions begins to take over. One question is how far an enzyme active site structure can be disrupted before it loses its capacity to approximate the behavior of the WT protein. The simple act of mutating pairs of hydrophobic side chains in SLO-1 has turned out to be distribution is allowed. distribution incredibly informative in this regard. New data have recently emerged for a double 546 754 mutant of SLO-1, Leu Ala/Leu Ala. A combination of steady-state and single turnover kinetics, X-ray crystallography and theoretical modeling on this variant 9180 provide a powerful test/conﬁrmation of the model outlined above for enzymatic only. No further No only. C–H activation [7]. Enzyme-Catalyzed Methyl Transfer Reactions. In order to ascertain the generality of reduced barrier width in native enzyme catalysis, studies have been extended to include the large family of methyl group-transferring enzymes. Methyl trans-

vayinstitutes.be/ fer is as ubiquitous in Nature as C–H activation, occurring at the levels of small PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- molecule, protein, RNA and DNA modiﬁcation. To initiate our studies, we chose an enzyme with well-established structural and kinetic characterizations, human catechol-O-methyltransferase (COMT) in which a methyl group is transferred from http://www.sol S-adenosylmethionine (AdoMet) to the oxygen of a catecholate. Very early studies on this enzyme compared the magnitude of secondary KIEs, host on on host using AdoMet labeled with deuterium at the reactive methyl group, to a small molecule model reaction. A moreinversevaluefortheKIEintheenzymereaction led to the proposal of catalysis via transition state compression [8]. This was a unique and controversial ﬁnding that stimulated considerable activity among theo-

http://www.worldscientific.com/worldscibooks/10.1142/ reticians. Following on our studies of C–H tunneling, that show control of active site compaction by bulky amino acid side chains, we pursued site-speciﬁc mutagenesis in COMT. The position targeted was Tyr68, a conserved residue that sits behind the sulfonium atom of AdoMet. From inspection of X-ray structures of COMT, key interactions involving Tyr68 include a close approach (within van der Waals contact) of its β carbon and the sulfonium center of cofactor, as well as an inferred hydrogen bonding interaction between its ring hydroxyl group and a nearby Glu. Kinetic

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World studies were carried out using cofactor labeled with three tritiums at the activated

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY methyl group, to increase the sensitivity of measurements beyond what is possible with deuterated cofactor [9]. The results indicate a trend in both kcat and kcat/KM as Tyr68 is converted to either Phe or Ala; substitution of WT (108V) by a variant that maps to documented disease states (V108M) gives similar results. A plot of kcat/KM for reaction of the preformed COMT-AdoMet complex with dopamine (Fig. 2) shows a remarkable linear correlation between the size of the second order October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 68

68 J. P. Klinman

rate constant and the magnitude of the secondary KIE. One explanation for the eﬀect, oﬀered at the time of publication, was a transition state compaction in the native protein that is substantially lost on reduction in the size of the Tyr68 behind

the methyl group donor. These studies have recently been extended, with the goal of

establishing a molecular basis for the correlation shown in Fig. 2. A combination of experimental binding isotope eﬀect measurements, high-level GPU-based QM/MM calculations that allow up to 30% of COMT to be analyzed using ab initio methods [10] and classical MD simulations point toward an important role for ground state compaction in COMT [11]. distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

http://www.sol Fig. 2. Linear relationship between kcat/KM(dopamine) and the magnitude of the 2 KIE in COMT [9]. host on on host Perspectives The studies outlined above implicate a general role for active site compaction in enzyme catalysis. This can be achieved by a sampling of protein conformational sub-

http://www.worldscientific.com/worldscibooks/10.1142/ states (H-tunneling) or through a combination of structural and dynamical eﬀects (methyl transfer). The methyl group transfer study extends the concept of active site compaction beyond the realm of systems that are dominated by quantum mechanical tunneling. For the future, it will be extremely important to pursue studies of a third enzymatic group transfer reaction. Phosphoryl transfer is a very attractive target, given its central role in biology. This will require developing a set of tools that can assess compaction and its sensitivity to alteration in protein struc- ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World ture/dynamics. The presence of active site compaction implicates a role for protein

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY structure that goes beyond the “inner” active site region. Novel protein engineering approaches, that allow investigation of the role of overall protein size in catalysis, may be of considerable interest.

Acknowledgments This work has been supported by the NIH (GM025765) and NSF (MCB0446395). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 69

Looking in New Directions for the Origins of Enzymatic Rate Accelerations 69

References 1. R. Wolfenden, Chem. Rev. 106, 3379 (2006); D. R. Edwards, D. C. Lohman, R. Wolfenden, J. Amer. Chem. Soc. 134, 525 (2012).

2. S. J. Benkovic, G. G. Hammes, S. Hammes-Schiﬀer, Biochemistry 47, 3317 (2008). 3. S. Neidle, R. K. Allemann, N. S. Scrutton (eds.), Quantum Tunnelling in Enzyme-Catalysed Reactions, RSC Biomolecular Sciences, Cambridge, UK (2009). 4. J. P. Klinman, A. Kohen, Annu. Rev. Biochem. 82, 471 (2013). 5. R. A. Marcus, J. Chem. Phys. 24, 966 (1956). distribution is allowed. distribution 6. M. P. Meyer, D. R. Tomchick, J. P. Klinman, Proc. Natl. Acad. Sci. USA 105, 1146 (2008).

9180 7. S. Hu, S. C. Sharma, A. D. Scouros, A. V. Soudackov, C. A. M. Carr et al., J. Amer. Chem. Soc. 136, 8157 (2014). only. No further No only. 8. M. F. Hegazi, R. T. Borchardt, R. L. Schowen, J. Amer. Chem. Soc. 101, 4359 (1979). 9. J. Y. Zhang, J. P. Klinman, J. Amer. Chem. Soc. 133, 17134 (2011). 10. H. J. Kulik, J. Zhang, J. P. Klinman, T. J. Martinez, in preparation. vayinstitutes.be/ 11. J. Zhang, H. J. Kulik, T. J. Martinez, J. P. Klinman, in preparation. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 70

COMPUTATIONAL ENZYME DESIGN AND METHODS TO PREDICT THE ROLE OF REMOTE MUTATIONS

KENDALL N. HOUK Department of Chemistry and Biochemistry, University of California Los Angeles, CA 90095-1569, USA distribution is allowed. distribution

My view of the present state of research on exploring enzyme 9180 families and enzyme catalysis only. No further No only. The computational design of functional proteins will boost the practical applications of biocatalysis. For nearly ten years, my group has been involved in the development of methods for the computational design of new enzymes. With David Baker, Stephen Mayo, Donald Hilvert, and others we have applied quantum me-

vayinstitutes.be/ chanics, molecular dynamics, and combined bioinformatics/force-ﬁeld approaches PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- like Rosetta to predict protein sequences that will fold into functional enzymes. The consequences of a successful approach of this type will be enormous. The scientiﬁc impact of success will be the demonstration that we understand how en- http://www.sol zyme catalysts function at an atomistic level and have mastered the complexities of catalysis, enabling us to replace Nature’s random evolutionary screening by pre- host on on host diction based on ﬁrst-principles physics. More practically, catalysts, the mainstay of industrial synthesis, and even the proteins that enable and control metabolism, will be designed by computational prediction and then easily synthesized by bacteria programmed by DNA made automatically on DNA synthesizers. Indeed, groups

http://www.worldscientific.com/worldscibooks/10.1142/ around the world have successfully designed and redesigned functional proteins, but these accomplishments resulted from heroic eﬀorts, not routine protocols. The reality is that our success in designing enzymes is erratic at best, and usually only a small percentage of designs actually function properly. Methods to incorporate required functional groups into stable protein scaﬀolds have been devised, but directed evolution — not in general computationally predicted — often must be used to increase activity to useful levels. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World In order to improve computational design methods, we have undertaken the

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY computational investigation of the results of directed evolution. Our goal is to learn how to predict these mutations in advance of experiment, and to improve the methods for enzyme design.

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 71

Computational Enzyme Design and Methods to Predict the Role of Remote Mutations 71

My recent research contributions to exploring enzyme families and enzyme catalysis General work in the computational enzyme design ﬁeld has been reviewed [1]. The

general inside-out scheme we devised with David Baker is shown below in Fig. 1. Computational enzyme design is a classic multi-scale problem, using high-accuracy quantum mechanics to position catalytic groups precisely around the transition states of the reaction and more approximate quantum, force-ﬁeld and informatics methods for treatment of protein motions and solvation. First, quantum mechanical calculations are used to design the active site to catalyze reactions of entirely new mechanisms or unnatural substrates by providing distribution is allowed. distribution stabilizing interactions that lower the free energy of the transition state. We use motifs that Nature has evolved for diﬀerent types of reactions, but quantum mechan-

9180 ics are used to determine optimal positioning of catalytic groups for the transition state of the reaction of interest. We have shown, in collaboration with Hilvert, that only. No further No only. evolved active sites (e.g., of esterases) can be predicted by quantum mechanical op- timizations [2]. The computationally predicted three-dimensional arrangement of catalytic groups to stabilize a given transition state is called a theoretical enzyme vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- (a)

http://www.sol

host on on host (b)

http://www.worldscientific.com/worldscibooks/10.1142/

(c) ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Fig. 1. The Inside-Out Protocol for enzyme design. The panels represent (a) the quantum mechanical prediction of a theozyme (here for the Kemp elimination), (b) the incorporation of a theozyme into a stable protein, and (c). the redesign of the non-catalytic residues in the active site. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 72

72 K. N. Houk

or “theozyme” [1]. This theozyme is then incorporated into a stable protein of known structure in place of residues in the active site of the natural protein using the program from Baker’s group, RosettaMatch (known as Matcher in the current

implementation). Remaining side-chains that pack the active site are redesigned

with RosettaDesign (known as EnzDes in the current implementation). Our group has also developed SABER [3] to search crystal structures of proteins in the PDB and to locate enzymes that already have the necessary arrangements of functional groups. The designs produced from Rosetta or SABER are then evaluated according to Rosetta folding criteria and by comparison to theozyme geometries. The best distribution is allowed. distribution are then subjected to molecular dynamics simulations in the presence of substrates,

counterions, and water [4]. When MD causes the active site to deviate substantially from that predicted by QM to be required for catalysis, then that design is discarded. 9180 This procedure has been successful for the design of Kemp eliminases, retro- only. No further No only. aldolases, and stereospeciﬁc Diels–Alderases, plus some others. Variations on this approach have been successful in our collaborations with Mayo [5]. The role of molecular dynamics in designs has been demonstrated for many cases [6]. Our recent studies have turned to perfecting the post-Rosetta or post-SABER

vayinstitutes.be/ redesign process. We wish to computationally predict what now is only achieved PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- experimentally by directed evolution [7]. Tolearnhowtodothis,weareevaluat- ing the pathways of successful directed evolution experiments on several successful commercial enzymes. Our current eﬀorts involve the theoretical study of enzymes http://www.sol evolved by our colleagues Yi Tang and Todd Yeates at UCLA [8] and by Gjalt Huis- man’s group at Codexis [9], to produce enzymes that can be used for the commercial host on on host production of drugs like sitagliptin, simvastatin, or montelukast. One example of this involves a protein, LovD, that naturally accepts an alcohol to produce the drug lovastatin [8, 9]. LovD uses an acyl carrier protein to deliver the acyl group. Tang and Codexis evolved LovD to transfer a modiﬁed acyl group

http://www.worldscientific.com/worldscibooks/10.1142/ from a much smaller carrier. Figure 2 summarizes the results of our work on this evolutionary path [10], involving many long-timescale molecular dynamics (MD) runs on the special-purpose computer, ANTON [11]. Panel A shows an arrangement of functional groups in the catalytic triad that is ineﬀective in catalysis, while B shows an arrangement that is good for catalysis. The MD studies showed that remote mutations gradually produce catalytically potent arrangements of the catalytic triad. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Outlook to future developments of research on exploring enzyme families and enzyme catalysis

Armed with information on how remote mutations influence pre-organization of active sites, we are devising computational protocols to build such information into the design process. The first step in this is to use iterative Rosetta/MD [12] to monitor how remote mutations influence critical catalytic distances. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 73

Computational Enzyme Design and Methods to Predict the Role of Remote Mutations 73

(a) (c)

(b) distribution is allowed. distribution

9180 only. No further No only. Fig. 2. Catalytically inactive (a) and active (b) arrangements of the Ser76–Lys79–Tyr188 triad in evolved LovD variants and protein-protein complexes revealed by MD simulations. (c) The catalytic Tyr188–Lys79 distance along the MD trajectories. Catalytic and non-catalytic regimes have been depicted in green and red, respectively. The less active mutants involve distances mostly in the red zone — signaling inactivity. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- The fields of enzyme design and directed evolution are advancing rapidly. Simul- taneously, computational methods — quantum mechanical and force-field-based, are being made more accurate and efficient, and the inexorable increases in computer http://www.sol power make larger and better calculations feasible. The importance of designed enzymes as catalysts will expand rapidly in research, commerce and medicine. host on on host

Acknowledgments Our research has been supported by the Defense Advanced Research Projects

http://www.worldscientific.com/worldscibooks/10.1142/ Agency, the National Institute of General Medical Sciences, National Institutes of Health, and Maersk Oil Research and Technology Center. Computer services from XSEDE and PSC (ANTON) were provided by the National Science Foundation.

References 1. G. Kiss, N. Celebi-Olcum, R. Moretti, D. Baker, K. N. Houk, Angew. Chem.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Int. Ed. 52, 5700 (2013). 2. A. J. T. Smith, R. Muller, M. D. Toscano, P. Kast, H. Hellinga et al., J. Amer. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Chem. Soc. 130, 15361 (2008). 3. G. R. Nosrati, K. N. Houk, Prot. Sci. 21, 697 (2012).

4. G. Kiss, D. Rothlisberger, D. Baker, K. N. Houk, Prot. Sci. 19, 1760 (2010). 5. H. K. Privett, G. Kiss, T. M. Lee, R. Blomberg, R. A. Chica et al., Proc. Natl. Acad. Sci. 109, 3790 (2012). 6. G. Kiss, V. S. Pande, K. N. Houk, Meth. Enzymol. 523, 145 (2013). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 74

74 K. N. Houk

7. U. T. Bornscheuer, G. W. Huisman, R. J. Kazlauskas, S. Lutz, J. C. Moore et al., Nature 485, 185 (2012). 8. X. Gao, X. Xie, I. Ashkov, J. R. Sawaya, J. Laidman et al., Chem. Biol. 16,

1064 (2009).

9. R. J. Fox, S. C. Davis, E. C. Mundorff, L. M. Newman, V. Gabrilovic et al., Nat. Biotech. 25, 338 (2007). 10. G. Jiménez-Osés,S.Osuna,X.Gao,M.R.Sawaya,L.Gilsonet al., Nat. Chem. Biol. 10, 431 (2014). 11. R. O. Dror, R. M. Dirks, J. P. Grossman, H. Xu, D. E. Shaw, Annu. Rev. Biophys. 41, 429 (2012). distribution is allowed. distribution 12. S. Lindert, J. Meiler, J. A. McCammon, J. Chem. Th. Comput. 9, 3843 (2013).

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 75

DISCOVERING NOVEL ENZYMES, METABOLITES AND PATHWAYS

JOHN A. GERLT Institute for Genomic Biology, University of Illinois, Urbana-Champaign 1206 West Gregory Drive, Urbana, IL 61801, USA distribution is allowed. distribution

My view of the present state of research on “exploring enzyme 9180 families and enzyme catalysis” only. No further No only. Homologous enzymes can catalyze the same overall reaction (isofunctional enzyme family), the same chemical reaction using different substrates (specificity diverse enzyme family), or different overall reactions that share a conserved partial reaction (mechanistically diverse enzyme superfamily). Before genome sequencing was

vayinstitutes.be/ routine, the members of enzyme (super)families were discovered by experimental PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- characterization. However, now the members of enzyme (super)families are discovered without regard to either in vitro activity or in vivo metabolic/physiological function. As of August 2013, the nonredundant UniProtKB/TrEMBL database http://www.sol contained 42,821,879 sequences; this number is increasing rapidly. However, a conservative estimate is that ≥ 50% of these sequences have uncertain, unknown, or host on on host incorrectly annotated functions. Indeed, as of August 2013, only 540,958 sequences had human-curated functions in the SwissProt database. Without high-throughput strategies and tools to decipher the functions of enzymes discovered in genome projects, the biomedical, industrial, and systems biology potential of these projects

http://www.worldscientific.com/worldscibooks/10.1142/ cannot be realized.

My recent research contributions to “exploring enzyme families and enzyme catalysis” The Enzyme Function Initiative (EFI) [1], a large-scale collaborative project supported by the National Institute of Health, is addressing the challenge of identify-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World ing the in vitro activities and in vivo metabolic/physiological functions of enzymes discovered in genome projects. The EFI brings together expertise in bioinformat- NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY ics, protein production and structure determination, homology modeling, in silico ligand docking [2, 3], experimental enzymology, genetics, transcriptomics, and

metabolomics to devise tools and strategies that can be used to predict and verify novel enzymatic activities and metabolic pathways encoded by microbial genomes. The recognition of the importance of the human microbiome in human health jus- tiﬁes the EFI’s focus on microbial enzymes, although many of the strategies and tools can be used to discover the functions of uncharacterized eukaryotic enzymes. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 76

76 J. A. Gerlt

As an example of this approach [4], the EFI recently “deorphanized” an enzyme from the pelagic bacterium Pelagibaca bermudensis HTCC2601 for which the New York Structural Genomics Research Consortium (a high-throughput structural

genomics project) had determined a high-resolution X-ray structure (PDB Code

2PMQ). The protein, now designated HpbD, is a member of the mechanistically diverse enolase superfamily [5]. The gene encoding HpbD (hpbD) is located in gene cluster that could encode a metabolic pathway (Fig. 1). distribution is allowed. distribution

9180 Fig. 1. Genome neighborhood context of the gene encoding 2PMQ (hpbD). only. No further No only.

Homology models were constructed for both a periplasmic solute binding protein component of an ABC transport system (HpbJ) and a dioxygenase (HpbB1); in silico docking of virtual ligand libraries to all three proteins (“pathway docking” vayinstitutes.be/ [6]) was used to predict metabolites that bind to the active sites of HpbD, HpbJ, and PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- HpbB1. Taken together, the results allowed the prediction that HpbD binds trans- 4-hydroxy-L-proline betaine (tHyp-B) and catalyzes its 2-epimerization to cis-4-

http://www.sol hydroxy-D-proline betaine (cHyp-B). Subsequent enzymatic activity measurements −1 −1 −1 conﬁrmed this prediction (kcat, 330 ± 30 sec ;kcat/KM, 7800 M sec ). A high-

host on on host resolution X-ray structure conﬁrmed the validity of the predicted pose of tHyp-B in the active site of HpbD. Thus, the 4-hydroxyproline 2-epimerase function can be assigned to HpbD and its orthologues in the databases (Fig. 2). http://www.worldscientific.com/worldscibooks/10.1142/

Fig. 2. Reaction catalyzed by HpbD. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY The family/superfamily memberships of the enzymes encoded by the neighboring genes allowed the prediction of a catabolic pathway for tHyp-B (Fig. 3):

successive oxidative removal of both betaine methyl groups to yield cHyp which is oxidized to its imino acid (by a member of the D-amino acid oxidase family), dehydrated and “hydrolyzed” to α-ketoglutarate semialdehyde (by a member of the dihydrodipicolinate synthase superfamily), and, ﬁnally, oxidized to α-ketoglutarate October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 77

Discovering Novel Enzymes, Metabolites and Pathways 77

distribution is allowed. distribution

9180 Fig. 3. Pathway for catabolism of tHyp-B. only. No further No only.

(by a member of the aldehyde dehydrogenase family). This pathway was conﬁrmed by metabolomics and genetics. The in vivo physiological function of an orthologue of HpbD encoded by Para-

vayinstitutes.be/ coccus denitriﬁcans PD1222 was determined by microbiology and transcriptomics PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- (in contrast to P. bermudensis, P. denitrificans PD1222 can be grown in the laboratory and is genetically tractable). Betaines often are used by bacteria as osmolytes to reduce stress by the presence of high salt in the growth medium. The growth http://www.sol of P. denitrificans is inhibited by 0.5 M NaCl in glucose medium; this inhibition is alleviated by tHyp-B, establishing that tHyp-B is an osmolyte. The genes encoding host on on host transporters for tHyp-B are up-regulated by tHyp-B and cHyp-B, thereby allowing their accumulation (with 20 mM tHyp-B in the growth medium, the intracellular concentration of Hyp-B is 170 mM). However, the genes encoding HpbD as well as all of the downstream enzymes in the catabolic pathway are down-regulated by http://www.worldscientific.com/worldscibooks/10.1142/ tHyp-B and cHyp-B; thus, the flux of tHyp-B through the catabolic pathway is regulated so that it is not catabolized when it is needed as an osmoprotectant. −1 −1 The “modest” value of the kcat/KM, 7800 M sec , for HpbD is physiologically relevant because of the multi-millimolar concentration of its substrate. This example illustrates how multidisciplinary collaboration allows prediction and verification of in vitro activities and in vivo metabolic/physiological functions of uncharacterized enzymes discovered in genome projects. The EFI continues to ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World develop new bioinformatic and computational tools to enable the biological com-

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY munity to adopt these strategies.

Outlook to future developments of research on “exploring enzyme

families and enzyme catalysis” The challenge of determining the in vitro activities and in vivo metabolic/ physiological functions of uncharacterized enzymes discovered in genome projects October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 78

78 J. A. Gerlt

will increase for the foreseeable future — the number of microbial species that are important for human health, bioremediation, and natural product isolation will continue to increase as new environmental niches are explored. Computational tools,

including sequence similarity networks [7], homology modeling, and in silico ligand

docking, will continue to be developed and reﬁned to better meeting this challenge. These tools and the strategies with which they are applied must be disseminated to and used by the community.

Acknowledgment distribution is allowed. distribution

The EFI is supported by the National Institutes of Health (U54GM093342).

9180 References 1. J. A. Gerlt, K. N. Allen, S. C. Almo, R. N. Armstrong, P. C. Babbitt et al., Bio- only. No further No only. chemistry 50, 9950 (2011). 2.J.C.Hermann,R.Marti-Arbona,A.A.Fedorov,E.Fedorov,S.C.Almo et al., Nature 448, 775 (2007). 3. L. Song, C. Kalyanaraman, A. A. Fedorov, E. V. Fedorov, M. E. Glasner vayinstitutes.be/ et al., Nat. Chem. Biol. 3, 486 (2007). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 4. S. Zhao, R. Kumar, A. Sakai, M. W. Vetting, B. M. Wood et al., Nature 502, 698 (2013).

http://www.sol 5. J. A. Gerlt, P. C. Babbitt, M. P. Jacobson, S. C. Almo, J. Biol. Chem. 287,29 (2012).

host on on host 6. C. Kalyanaraman, M. P. Jacobson, Biochemistry 49, 4003 (2010). 7. H. J. Atkinson, J. H. Morris, T. E. Ferrin, P. C. Babbitt, PLoS One 4, 34345 (2009). http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 79

PROGRAMMING NEW CHEMISTRY INTO THE GENETIC CODE OF CELLS AND ANIMALS

JASON W. CHIN Centre for Chemical and Synthetic Biology, Division of Protein and Nucleic Acid Chemistry Medical Research Council Laboratory of Molecular Biology, Francis Crick Avenue Cambridge CB2 0QH, UK distribution is allowed. distribution

9180 My view of the present state of research on exploring enzyme families and enzyme catalysis only. No further No only.

There has been rapid progress in developing approaches to incorporate unnatural amino acids into proteins. Early approaches used chemical aminoacylation of tRNAs in in vitro translation systems. Progress in this area has been facilitated

vayinstitutes.be/ by the development of RNA based aminoacylation systems that can be used in PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

vitro as well as puriﬁed translation systems in which the ribosomes are isolated - from cells but other translation components are made recombinantly. In contrast to cellular translation, in vitro translation systems commonly use stoichiometri-

http://www.sol cally pre-aminoacylated tRNAs that are not- re-aminoacylated in the translation mix. It has been possible to select unnatural peptides and cyclic peptides bearing

host on on host several unnatural monomers using in vitro translation and selection approaches and in vitro translation approaches have been key is clarifying the chemical scope of the ribosome for unnatural monomers. Methods for incorporation unnatural amino acids into proteins in vivo promise

http://www.worldscientific.com/worldscibooks/10.1142/ scalable synthesis of modiﬁed proteins and provide the basis of powerful approaches for studying biological phenomena. Two main approaches for unnatural amino acid incorporation in cells have been described. One approach takes advantage of the permissivity of existing cellular aminoacyl-tRNA synthetases (or their active site mutants) to incorporate a structurally limited set of unnatural amino acids in response to sense codons. Typically these unnatural amino acids are incorporated sub-stoichiometrically creating a statistical mixture of proteins, though experiments

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World in auxotrophic strains allow near quantitative incorporation of some amino acids in response to a sense codon in a target protein. These approaches have been NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY useful for pulse-labeling the proteomes of cells. A distinct approach introduces orthogonal translational components and aims to genetically encode the quantitative

installation of unnatural amino acids in response to a deﬁned blank codon(s), most commonly the amber stop codon. An orthogonal ribosome has been created and evolved to decode new bank codons (including quadruplet codons), facilitating the incorporation of multiple distinct unnatural amino acids in a single polypeptide. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 80

80 J. W. Chin

Orthogonal aminoacyl-tRNA synthetase/tRNA pairs (most-notably the pyrrolysyl- tRNA synthetase/tRNA pair) have been developed for the incorporation of diverse amino acids. Genetic code expansion has been developed in E. coli,yeast,mam-

malian cells and most recently in animals: C. elegans and D. melanogaster.The

chemical scope of modiﬁcations has been augmented by combining genetic code expansion and chemoselective modiﬁcation. The approaches developed have been used to obtain previously unattainable insights into the molecular basis of biological function, both by the synthesis of recombinant proteins for in vitro structure function studies and by developing approaches for controlling and imaging biological processes in cells and animals. distribution is allowed. distribution

My recent research contributions to exploring enzyme families and

9180 enzyme catalysis only. No further No only.

Reprogramming Cellular Translation: In the cell, DNA is copied to messenger RNA, and triplet codons (64) in the messenger RNA are decoded — in the process of translation — to synthesize polymers of the natural 20 amino acids. This process (DNA-RNA-protein) describes the central dogma of molecular biology and is con-

vayinstitutes.be/ served in terrestrial life. We are interested in re-writing the central dogma to create PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

organisms that synthesize proteins containing unnatural amino acids and polymers - composed of monomer building blocks beyond the 20 natural amino acids. We have invented and synthetically evolved new ‘orthogonal’ translational components

http://www.sol (including ribosomes [1] and aminoacyl-tRNA synthetases) and demonstrated the assembly of these components into an ‘orthogonal translation pathway’ to address

host on on host major challenges in re-writing the central dogma of biology in cells. The approaches we have developed address the outstanding problem of altering the function of conserved cellular hubs in cells that orchestrate essential functions (in this case the ribosome). We have shown that by creating orthogonal (and non-essential) versions

http://www.worldscientific.com/worldscibooks/10.1142/ of these hubs it is possible to evolve their molecular function. We have evolved the orthogonal ribosome to read the amber stop codon as a sense codon and to read quadruplet codons [2, 3]. Using these ribosomes that have additional decoding capacity in combination with mutually orthogonal synthetase/tRNA pairs we have demonstrated the ﬁrst encoded incorporation of multiple useful unnatural amino acids into protein in cells [3]. By programming two amino acids that can react together into proteins we have demonstrated the genetically directed proximity ac-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World celeration of a cycloaddition reaction on a protein to cyclize a protein [3]. Developing the Pyrrolysyl tRNA synthetase tRNA pair for unnatural amino acid NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY incorporation in cells and animals: We have extended the foundational technologies we have pioneered to develop new approaches to address unmet challenges in

understanding biological processes with spatial, temporal and molecular precision [4]. Much of this work is underpinned by our development of the pyrrolysyl-tRNA synthetase/tRNACUA pair for genetic code expansion. We have demonstrated that this pair, unlike the previous aminoacyl-tRNA synthetase/tRNA pairs that have October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 81

Programming New Chemistry into the Genetic Code of Cells and Animals 81

been used for genetic code expansion, is orthogonal in E. coli, S. cerevisiae,mammalian cells, C. elegans and D. melanogaster [5–9] and this pair can be used to incorporate a range of aliphatic and aromatic amino acids. We have also demon-

strated that this pair can be evolved in E. coli to speciﬁcally incorporate unnatural

amino acids [5]. These advances have rapidly expanded the scope and accessibility of genetic code expansion because they allows aminoacyl-tRNA synthetases developed for the incorporation of new amino acids in E. coli to be transplanted and used for unnatural amino acid incorporation in eukaryotic cells and animals. Insights into the molecular consequences of post-translational modifications via genetic code expansion: A catalog of post-translational modifications have been distribution is allowed. distribution identified in proteins, and mass spectrometry based approaches are rapidly expand-

ing the number of modifications identified. One way to provide insight into the roles of modifications in regulating protein structure and function is to make and study 9180 the modified protein. However, in many cases there is no way to make the modified only. No further No only. proteins for structural and functional studies because the modifying enzymes are unknown or do not modify a given site specifically or quantitatively in vitro.We have developed approaches for genetically encoding post-translational modifications (including lysine acetylation [5] and mono- and di- methylation [4]), as well as the

vayinstitutes.be/ development of approaches that combine chemical synthesis and protein chemistry PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- with genetic code expansion to allow the genetically directed installation of ubiquitination [10, 11], a post-translational modification that is challenging to install by either chemistry or genetic code expansion alone. http://www.sol We have applied our approaches to installing post-translational modifications into recombinant proteins to obtain previously unattainable insight into biological host on on host regulation through interfacing our approaches with structural biology, enzymology and single molecule methods. Using these approaches we have defined the effects of a key acetylation on chromatin in regulating the wrapping of DNA around nucleo- somes [4], the effect of cyclophilin acetylation in regulating cis-trans isomerization

http://www.worldscientific.com/worldscibooks/10.1142/ of HIV capsid and mediating cyclosporine binding (effects that may be important in immunosuppression and viral infection) [4], synthesized atypical ubiquitin linkages for the first time, determined the first structure of K6-linked ubiquitin linkages and profiled a variety of deubiquitinases on atypical ubiquitin linkages for the first time [4]. Our approaches for synthesizing post-translationally modified proteins have been widely adopted by many other groups and been used to provide additional insights into diverse biological processes.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Imaging biological processes in cells and organisms: Understanding how proteins

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY function inside cells will be facilitated by approaches that allow proteins to be labeled and imaged with small molecule fluorophores. We have developed approaches for the rapid and site-specific labelling of proteins in cells with small molecule fluorophores. These approaches involve genetically encoding an unnatural amino acid at a defined site in a protein of interest and labeling this site with a fluorophore or other small molecule. Key to this strategy is the development and encoding of func- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 82

82 J. W. Chin

tional groups that (i) are bioorthogonal and (ii) can be labeled with probes in rapid reactions. We have demonstrated the encoding of strained alkenes and alkynes and their labeling with tetrazine based probes in ﬂuorogenic inverse electron demand

Diels-Alder reactions that produce nitrogen gas as the sole byproduct [12, 13]. This 5 −1 −1

reaction (rate constants upto 10 M s ) is 4-5 orders of magnitude faster than most previous bioorthogonal reactions reported, and this approach has allowed the labeling of proteins at genetically encoded unnatural amino acids in and on mammalian cells for the first time. The minimally perturbing imaging approaches we have developed allow imaging of proteins in cells, without large fluorescent tags and with very bright fluorophores. This is facilitating super-resolution imaging of distribution is allowed. distribution proteins and other investigations into protein function.

Temporal dissection of signaling: Understanding the kinetics of elementary steps in signaling and the role of feedback and feedforward control in regulating cellular 9180 signals is an important challenge. We have demonstrated that by genetically en- only. No further No only. coding a photocaged lysine [7, 14] in place of a near universally conserved lysine in the active site of protein kinases (and introducing mutations that would make the wild-type kinase constitutively active) it is possible to create kinases in cells that can be activated by a pulse of light. This allows the activation of kinase signaling

vayinstitutes.be/ from any point in a cascade. This approach is allowing the dissection of the con- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- tribution of each step in a signaling cascade to signal propagation and adaptation [14]. Expanding the genetic code of animals: We have recently developed genetic code http://www.sol expansion approaches in whole animals (C. elegans and D. melanogaster) [8, 9] for the ﬁrst time. One goal of this work is to develop and apply the tools we have host on on host created for manipulating and imaging processes in cell culture to biological processes including development, disease progression, and cognitive function best studies at the level of whole organisms.

http://www.worldscientific.com/worldscibooks/10.1142/ Outlook to future developments of research on exploring enzyme families and enzyme catalysis The ability to alter the functions of some of the most central and conserved biological systems and redirect these systems in cells to new ends provides a powerful paradigm for both the study of biology and the discovery and manufacture of new therapeutics and materials. The approaches we are developing to genetically encode unnatural

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World amino acids into proteins in cells and animals are providing the foundation for a suite of new approaches to studying complex biological phenomena with molecular, NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY spatial and temporal precision. By combining the chemist’s ability to precisely tailor the properties of small molecules with redesigned biology that places these

small molecules at precise positions in proteins in cells we can maximally leverage the unique advantages of chemistry and biology for synthesis and discovery. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 83

Programming New Chemistry into the Genetic Code of Cells and Animals 83

Acknowledgments Work in our laboratory is funded by the Medical Research Council, The European Research Council, The Human Frontiers of Science Programme, and The Louis-

Jeantet Foundation.

References 1. O. Rackham, J. W. Chin, Nat. Chem. Biol. 1, 159 (2005). 2. K. Wang, H. Neumann, S. Y. Peak-Chew, J. W. Chin, Nat. Biotechnol. 25, 770 (2007). distribution is allowed. distribution 3. H. Neumann, K. Wang, L. Davis, M. Garcia-Alai, J. W. Chin, Nature 464, 441

(2010).

9180 4. L. Davis, J. W. Chin, Nat. Rev. Mol. Cell. Biol. 13, 168 (2012). 5. H. Neumann, S. Y. Peak-Chew, J. W. Chin, Nat. Chem. Biol. 4, 232 (2008). only. No further No only. 6.S.M.Hancock,R.Uprety,A.Deiters,J.W.Chin,J. Amer. Chem. Soc. 132, 14819 (2010). 7. A. Gautier, D. P. Nguyen, H. Lusic, W. An, A. Deiters et al., J. Amer. Chem. Soc. 132, 4086 (2010).

vayinstitutes.be/ 8. S. Greiss, J. W. Chin, J. Amer. Chem. Soc. 133, 14196 (2011). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 9. A. Bianco, F. M. Townsley, S. Greiss, K. Lang, J. W. Chin, Nat. Chem. Biol. 8, 748 (2012). 10. S. Virdee, P. B. Kapadnis, T. Elliott, K. Lang, J. Madrzak et al., J. Amer. http://www.sol Chem. Soc. 133, 10708 (2011). 11. S. Virdee, Y. Ye, D. P. Nguyen, D. Komander, J. W. Chin, Nat. Chem. Biol. host on on host 6, 750 (2010). 12. K. Lang, L. Davis, J. Torres-Kolbus, C. Chou, A. Deiters et al., Nat. Chem. 4, 298 (2012). 13.K.Lang,L.Davis,S.Wallace,M.Mahesh,D.J.Coxet al., J. Amer. Chem.

http://www.worldscientific.com/worldscibooks/10.1142/ Soc. 134, 10317 (2012). 14. A. Gautier, A. Deiters, J. W. Chin, J. Amer. Chem. Soc. 133, 2124 (2011). ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 84

EXPANDING THE ENZYME UNIVERSE THROUGH A MARRIAGE OF CHEMISTRY AND EVOLUTION

FRANCES H. ARNOLD Division of Chemistry and Chemical Engineering, California Institute of Technology 210-41 1200 E. California Blvd., Pasadena, CA 91125, USA distribution is allowed. distribution

Present state of research on expanding enzyme catalysis beyond 9180 Nature only. No further No only. For more than twenty years this laboratory has used directed evolution to modify enzymes. It is now widely accepted that directed evolution can change substrate speciﬁcity or reaction selectivity in desired ways, even if it sometimes remains dif- ﬁcult in practice. It is no longer surprising that enzymes adapt readily by accumu-

vayinstitutes.be/ lating beneﬁcial mutations. And why should it be, since this is how Nature tailors PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- them for myriad biological roles? More diﬃcult to grasp is how Nature discovers new enzyme functions, particularly new catalytic activities. We know that the biological world’s diverse catalytic repertoire is the product of evolution by natural http://www.sol selection, but we have little understanding of how Nature’s tinkering generates new functions. Sometimes we are (un)lucky enough to catch them in the act — e.g., host on on host the acquisition of antibiotic resistance or the ability to degrade man-made toxins. But for the vast majority of activities, the fossil record is nonexistent or too sparse to tell the molecular story. This leaves us without much guidance for evolving new enzymes in the laboratory. Consequently we are forced to contaminate our evolu-

http://www.worldscientific.com/worldscibooks/10.1142/ tion experiments with knowledge — e.g., computational design [1, 2] — in order to jumpstart the discovery process. If every bad catalyst could become a good one by directed evolution, part of the problem would be solved. We would be able to take any catalytic antibody or computationally-designed enzyme (or bovine serum albumin) and convert it into a great catalyst with multiple rounds of random mutagenesis and screening. We have learned the hard way, however, that not every bad catalyst lies at the base of ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World a tall ﬁtness peak, at least one that can be scaled by a random uphill walk. We

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY therefore want to know what features make a protein with a new catalytic activity the potential mother of a whole new enzyme family. And, what are good ways to ﬁnd new enzyme activities in the ﬁrst place?

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 85

Expanding the Enzyme Universe through a Marriage of Chemistry and Evolution 85

Recent research contributions to creating new enzymes My laboratory has been directing the evolution of a remarkable enzyme, a bacterial cytochrome P450, for at least a dozen years. This particular P450, from Bacillus

megaterium, is quite speciﬁc — it catalyzes subterminal hydroxylation of fatty acids. As a family, however, the P450s are wonderfully diverse, catalyzing a wide range of reactions on an even wider range of substrates. Nature has demonstrated the ‘evolvability’ of this bit of protein that binds an iron-heme (along with its various electron-transfer partners), and we spent several happy years creating new versions of the bacterial enzyme that could mimic known functions of other (e.g.,human) P450 family members [3, 4]. We could also extend its function beyond what was distribution is allowed. distribution known to be catalyzed by P450s, making, for example, versions that hydroxylate gaseous alkanes (propane and ethane), transformations thought to lie in the func-

9180 tional realm of the methane monooxygenase family [5]. The versatile iron-heme prosthetic group has been adopted by numerous trans- only. No further No only. port, signaling, and catalytic proteins. Within just the P450 family the range of reactions catalyzed is impressive: hydroxylation, epoxidation, sulfoxidation, per- oxidation, N- and O-dealkylation, and more. We found that we could access all of these reactivities, starting from wildtype or variants of our bacterial fatty acid vayinstitutes.be/ monooxygenase and using directed evolution to increase the initially-low activities. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- We learned that even a single P450 is highly evolvable and lies at the base of many different fitness peaks. The P450 is also catalytically promiscuous — the bacterial enzyme or its close variants could catalyze low levels of many different reactions. http://www.sol Presumably natural selection made the same discovery many times, setting old P450s to new tasks in many new contexts. host on on host More recently we decided to explore new reactions, not known in Nature. Re- actions whose mechanisms share features of the P450 machinery but have not been discovered in Nature’s many evolution experiments, because the context was not there. Understanding that natural selection can create new enzymes from promis- http://www.worldscientific.com/worldscibooks/10.1142/ cuous activities when presented with the opportunity to occupy a new niche (for example, a new substrate becomes available), we tested a collection of P450s for promiscuous activity in a few carbene and nitrene insertion reactions that are iso- electronic to the well-established formal ‘oxene’ transfer reaction of ferric-P450 enzymes with iodosylbenzene. The similarity is such that some of these reactions were explored in the 1980’s by ‘biomimetic’ chemists, notably Breslow and Dawson [6] and others [7]. The poor turnover numbers they reported for P450s, however, ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World discouraged further work, and no more was said for 30 years. But we knew from

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY experience that when a P450 is a bad catalyst, it can often easily become a good one. The wildtype P450 enzyme catalyzed just a few turnovers for cyclopropanation

of styrene with EDA, fewer even than free hemin (Table 1) and other heme proteins [8]. However, it was the only catalyst to exhibit enantioselectivity, indicating the reaction took place in an active site that could exert some control on selec- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 86

86 F. H. Arnold

tivity. We quickly identiﬁed variants in our collection, such as H2A10 and CIS (Table 1), that exhibited high selectivity for the cis diastereomer (opposite from that of hemin, which makes mostly the trans cyclopropane product), high activity,

and high enantioselectivity.

The big breakthrough came, however, when we replaced the cysteine residue that ligates the heme iron with Ser [9]. Our goal was to have the reaction proceed in vivo, where it would have to rely on endogenous NADPH rather than the sodium dithionite used to reduce the heme Fe(III) to Fe(II) in vitro. This generated a very active cyclopropanation enzyme that functions extremely well in vivo: the 67,800 turnovers for the P411-CIS enzyme (Table 1) is, we believe, the highest activity ever distribution is allowed. distribution reported for this reaction with any catalyst. The Cys-Ser mutation abolished all

monooxygenase activity and caused the typical peak at 450 nm in the CO-diﬀerence spectrum to shift to 411 nm. Thus we call this new catalyst a P411. Much more 9180 active than P450-CIS, the P411-CIS has a crystal structure nearly identical to that only. No further No only. of P450-CIS. We have started to diversify this enzyme by directed evolution to expand its substrate range and selectivity.

Table 1. Activities of diﬀerent P450 variants for styrene cyclopropanation [9]. Yields are based vayinstitutes.be/ on EDA. TTN = total turnover number. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host

Catalyst [EDA] [P450] % yield TTN cis:trans %eecis %eetrans (mM) (μM) http://www.worldscientific.com/worldscibooks/10.1142/ Hemin 10 20 15 73 6:94 1 0

P450BM3 10 20 1 5 37:63 27 2

9-10A TS F87V 10 20 1 7 35:65 41 8

H2A10 10 20 33 167 60:40 95 78

P450BM3-heme-CIS 8.5 15 32 212 77:23 94 91

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World P411BM3-heme-CIS 8.5 15 51 342 93:7 99 51

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY P450BM3-CIS* 8.5 3.7 42 950 22:78 60 22

P411BM3-CIS* 8.5 1.3 55 3700 76:24 96 25

BM3

P411 -CIS* 170 1.8 72 67800 90:10 99 43

∗Conducted with intact E. coli cells. # Conditions for reactions with puriﬁed P450s: 1 equiv Na2S2O4, Ar atmosphere, 0.1 M KPi pH 8.0. Conditions for reactions with intact E. coli cells: 0.2 equiv glucose, Ar atmosphere, M9-N medium. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 87

Expanding the Enzyme Universe through a Marriage of Chemistry and Evolution 87

Oleﬁn cyclopropanation is not (yet!) a biologically relevant transformation, because P450s do not encounter the reactive diazoesters in their native environments. They nonetheless have this promiscuous activity, which can be captured by evolu-

tion when the opportunity arises. Our work shows that this promiscuous activity

can be enhanced signiﬁcantly with just a few mutations, something that can happen readily in a protein evolving under selective pressure, either natural or forced. Enzymes that catalyze the concerted oxidative amination of C-H bonds are also apparently absent from Nature’s catalyst repertoire. Synthetic chemists, who are not limited to biologically accessible reagents and metals, have developed useful methods for C-H amination through a nitrenoid intermediate that also has no distribution is allowed. distribution parallel in natural enzymes. Following up on studies performed in the 1980’s [6],

we investigated whether our P450 and P411 enzymes could catalyze intramolecular C-H amination of aryl sulfonylazides to form benzosultams (example shown in 9180 Table 2) [10]. Whereas wildtype P450 showed only weak activity, some P411 vari- only. No further No only. ants catalyzed several hundred TTN. The puriﬁed enzymes as well as intact E. coli cells expressing the enzymes catalyze the amination reaction under anaerobic conditions (Table 2) [10]. P411- T268A and P411-CIS exhibited good activity and reasonable enantioselectivity vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Table 2. P450 and P411 enzymes catalyze direct C-H amination. Comparison of total turnover numbers (TTN) and enantioselectivities of intact E. coli cells expressing P450 and P411 variants, with azide 1 at 0.1 mol% catalyst loading, giving http://www.sol sulfonamide 2 and benzosultam 3 [10]. host on on host http://www.worldscientific.com/worldscibooks/10.1142/ In vivo catalyst [P450] or Yield 3 [%] TTN[a] ee [%][b] [P411] [(μM)]

Empty vector 0 0 0 n.d.

P450BM3 6.6 0.5 5.1 n.d

P450BM3- 5.8 7.8 26 84

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World P411BM3 4.3 6.7 29 16 NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY P411BM3- 2.2 30 250 89

P411BM3-CIS 1.4 46 680 60

P411BM3-CIS-T438S 2.7 58 430 87

[a] TTN = total turnover number to benzosultam 3. [b] *(S−R)/(S+R). n.d. = not determined. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 88

88 F. H. Arnold

(up to 89%); adding the T438S mutation to P411-CIS increased enantioselectivity (430 TTN, 86% ee). Optimization of expression conditions increased the productiv- ity of whole-cell C-H amination, enabling conversions to 3 of up to 66% in small-scale

reactions; higher yields have since been achieved at preparative scale.

Outlook for future enzymes The tools are now in place to create enzymes that catalyze reactions not known in Nature. We have reported oleﬁn cyclopropanation and C-H amination catalyzed by P450s and P411s, and there is clearly opportunity for more useful reactions based

distribution is allowed. distribution on this system. The highly evolvable P450 is an excellent starting point, both for

discovering new enzyme-catalyzed reactions and for diversifying them through di-

rected evolution. But there are likely many more such enzymes waiting for the right 9180 substrates to come along. A tasteful mix of chemical intuition, computational design where appropriate, and evolution (to circumvent our near complete ignorance only. No further No only.

of the details of the sequence-function code for catalysis) will generate whole new families of genetically-encoded catalysts, greatly expanding the catalyst repertoire for biosynthesis and for organic synthesis. What is more, we will be able to observe the creation of new biological functions and follow the mechanisms by which vayinstitutes.be/ they arise and are diversiﬁed and optimized. The future will see Nature’s chemical PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- universe expand to include more of the clever chemistry of man.

http://www.sol Acknowledgments I have described the work of highly talented graduate students and postdocs: Pedro host on on host Coelho, Eric Brustad, John McIntosh, Jane Wang and Chris Farwell. Support was received from the Jacobs Institute for Molecular Engineering for Medicine at Caltech, and the Gordon and Betty Moore Foundation through Grant GBMF2809 to the Caltech Programmable Molecular Technology Initiative. http://www.worldscientific.com/worldscibooks/10.1142/

References 1. F. Richter, R. Blomberg, S. D. Khare, G. Kiss, A. P. Kuzin et al., J. Amer. Chem. Soc. 134, 16197 (2012). 2. L. Giger, S. Caner, R. Obexer, P. Kast, D. Baker et al., Nat. Chem. Biol. 19, 494 (2013).

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 3.M.Landwehr,L.Hochrein,C.R.Otey,A.Kasrayan,J.-E.B¨ackvall et al., J. Amer. Chem. Soc. 128, 6058 (2006). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 4. A. M. Sawayama, M. M. Y. Chen, P. Kulanthaivel, M-S. Kuo, H. Hemmerle et al., Chem. Eur. J. 15, 11723 (2009).

5.R.Fasan,M.M.Chen,N.C.Crook,F.H.Arnold,Angew. Chem. 46, 8414 (2007). 6. E. W. Svastis, J. H. Dawson, R. Breslow, S. H. Gellman, J. Amer. Chem. Soc. 107, 6427 (1985). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 89

Expanding the Enzyme Universe through a Marriage of Chemistry and Evolution 89

7. J. R. Wolf, C. G. Hamaker, J.-P. Djukic, T. Kodadek, L. K. Woo, J. Amer. Chem. Soc. 117, 9194 (1995). 8. P. S. Coelho, E. M. Brustad, A. Kannan, F. H. Arnold, Science 339, 307 (2013).

9. P.S.Coelho,Z.J.Wang,M.E.Ener,S.A.Baril,A.Kannanet al., Nat. Chem.

Biol. 9, 485 (2013). 10.J.A.McIntosh,P.S.Coelho,C.C.Farwell,Z.J.Wang,J.C.Lewis et al., Angew. Chem. Int. Ed. 52, 9309 (2013). distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 90

CONTROLLED RADICAL REACTIONS IN BIOLOGY AND THE IMPORTANCE OF METALLO-COFACTOR BIOSYNTHESIS

JOANNE STUBBE Department of Chemistry, Massachusetts Institute of Technology Cambridge, MA 02139, USA distribution is allowed. distribution

My view of the present state of research on exploring enzyme 9180 families and enzyme catalysis only. No further No only. Radicals in general are considered to be inherently unstable and reactive and in Biology are almost always associated with destruction. The biological literature is rife with descriptions of reactive oxygen and nitrogen species generated as the inadvertent consequences of our environment and as byproducts of metabolism that

vayinstitutes.be/ if left unrestrained, result in modiﬁed macromolecules including DNA and protein. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Nature has evolved small molecules (VitC, VitE) and enzymes that are able to inactivate the “bad” radicals directly or repair the damage they inflict. Radicals are thus vilified. One of the most exciting areas in enzyme catalysis for the next decade, http://www.sol in my opinion, is to understand how Nature has been able to harness the reactivity of radicals to carry out very difficult chemical transformations with exquisite host on on host specificity. Good radicals in biology play a central role in primary metabolism from making the building blocks required for DNA biosynthesis, to the light mediated oxidation of water to O2. With the availability of the thousands of genomes and the ability to analyze them using bioinformatics, there are predicted to be > 50,000

http://www.worldscientific.com/worldscibooks/10.1142/ reactions that involve free radical chemistry [1]. Thus deﬁning the underlying principles of radical based reactions and the mechanisms by which radicals are generated will be an exciting area for the next decade.

My recent research contributions to exploring enzyme families and enzyme catalysis

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World My lab has been investigating for more than 30 years, how “good” radicals in biology are generated, protected, and used to catalyze the conversion of nucleoside NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 5’-diphosphates to deoxynucleotides by ribonucleotide reductases (RNRs). RNRs have been classiﬁed based on their metallo-cofactor that oxidizes an active site

cysteine to a transient thiyl radical which then initiates nucleotide reduction in all classes. The class I RNRs are subdivided into Ia that use a diferric tyrosyl radical cofactor [2], Ib that use a dimanganese-tyrosyl radical cofactor [3, 4] and Ic, that use a Mn(IV)Fe(III) cofactor [5]. The class II and III RNRs use adenosylcobalamin and October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 91

Controlled Radical Reactions and Metallo-Cofactor Biosynthesis 91

S-adenosylmethionine and a [4Fe4S]1+ cluster, respectively, both of which generate a 5’-deoxyadenosyl radical that initiates catalysis through a thiyl radical [6]. The RNR-catalyzed reactions involve stable and transient protein radicals (Fig. 1) and nucleotide radical intermediates. One of these cofactors SAM-[4Fe4S]1+ (Fig. 2) is

now proposed to be involved in thousands of reactions including the conversion of a methylphosphonate to methane gas [7] and the methylation of an adenine in the A site of the ribosome causing resistance to antibiotics [8]. distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ Fig. 1. Good radicals in biology: both stable and transient protein radicals. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host

1+ http://www.worldscientific.com/worldscibooks/10.1142/ Fig. 2. Good radicals in biology: the radical SAM superfamily of enzymes use SAM and [4Fe4S] to generate 5’-deoxyadenosyl radical which initates a broad range of reactions in both prokaryotes and eukaryotes.

New tools have been essential in analysis of how the metallo-cofactors initiate these radical reactions directly on the substrate or on the protein. These include rapid freeze quench high ﬁeld EPR and ENDOR spectroscopies, transient absorp-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World tion spectroscopies and the incorporation of isotopically labeled amino acids and unnatural amino acids site-speciﬁcally into proteins [9]. Daily, the structures of NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY chemically and kinetically competent radicals are being revealed. The requirement for metallo-cofactors to initiate these reactions has now made clear the importance

of understanding the biosynthetic and repair pathways that control the metallation process [10]. For example, the biosynthesis of the dimanganese-tyrosyl radical •− cofactor in one class Ib RNR requires a “bad” radical, O2 as the oxidant [11]. Lessons learned from studies of RNR and other well studied enzymes such as lysine October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 92

92 J. Stubbe

amino mutase [12], have set the stage for many chemical generalizations that have and will continue to emerge.

Outlook to future developments of research on exploring enzyme

families and enzyme catalysis The good news is that there are many novel mechanisms yet to be unraveled. The bad news is that many of the proteins require metallo-cofactors that are poorly incorporated in recombinant hosts during expression. When studying metal-based reactions, homogeneity of reaction centers is essential given the low sensitivity of the

distribution is allowed. distribution biophysical methods available. While we have a number of tools to look for, exam-

ine, and characterize biologically generated radicals, more sensitive tools that can

be used in vitro and in vivo need to be invented. Identifying and understanding the 9180 factors that assist and control metal and oxidant delivery, requires further investigation. Despite the amazing progress that has been made in FeS cluster biosynthesis only. No further No only.

in the last decade [13], for example, the source of iron and small molecules involved in sensing iron, still remain unknown. Metal homeostasis and redox balance are intimately linked in the cell and are essential to understand. Finally, as educators, we need to do a better job incorporating radical-dependent reactions including electron vayinstitutes.be/ transfer and proton coupled electron transfer mechanisms and the role of metals in PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- biology into our introductory biochemistry courses.

http://www.sol Acknowledgment I am grateful to the NIH in the USA for supporting my research throughout my host on on host entire career.

References

http://www.worldscientific.com/worldscibooks/10.1142/ 1. H. J. Soﬁa, G. Chen, B. G. Hetzler, J. F. Reyes-Spindoloa, N. E. Millerm, Nucl. Acids Res. 29, 1097 (2001). 2. J. A. Cotruvo, J. Stubbe, Annu. Rev. Biochem. 80, 733 (2011). 3. J. A. Cotruvo Jr., J. Stubbe, Biochemistry 49, 1297 (2010). 4. N. Cox, H. Ogata, P. Stolle, E. Reijerse, G. Auling et al., J. Amer. Chem. Soc. 132, 11197 (2010). 5. W. Jiang, D. Yun, L. Saleh, E. W. Barr, G. Xing et al., Science 316, 1188

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World (2007). 6. J. Stubbe, W. A. van der Donk, Chem. Rev. 98, 705 (1998). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 7. S. S. Kamat, H. J. Williams, L. J. Dangott, M. Chakrabarti, F. M. Raushel, Nature 497, 132 (2013).

8. T. L. Grove, J. S. Benner, M. I. Radle, J. H. Ahlum, B. J. Landgraf et al., Science 332, 604 (2011). 9. E. Minnihan, D. G. Nocera, J. Stubbe, Acc. Chem. Res. 46, 2524 (2013). 10. J. A. Cotruvo, J. Stubbe, Metallomics 4, 1020 (2012). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 93

Controlled Radical Reactions and Metallo-Cofactor Biosynthesis 93

11. J. A. Cotruvo, T. A. Stich, R. D. Britt, J. Stubbe, J. Amer. Chem. Soc. 135, 4027 (2013). 12. P. A. Frey, A. D. Hegeman, G. H. Reed, Chem. Rev. 106, 3302 (2006).

13. R. Lill, Nature 460, 831 (2009).

distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 94

SESSION 2: EXPLORING ENZYME FAMILIES AND ENZYME CATALYSIS

CHAIR: DONALD HILVERT AUDITORS: J.-M. FRERE` 1,W.VERSEES´ 2 (1) UniversitédeLiège, Centre for Protein Engineering, Institute of Chemistry, B6 Sart-Tilman, 4000 Liège, Belgium distribution is allowed. distribution (2) Structural Biology Brussels, Vrije Universiteit Brussel and Structural Biology

Research Center, VIB, Pleinlaan 2, 1050 Brussel, Belgium 9180

only. No further No only. Discussion among panel members

Don Hilvert to Judith Klinman: I think the idea of compaction in catalysis is quite compelling. In fact, this is what we saw in the evolution of our enzymes. All the groups came together, yielding a very precise alignment. However, why did you

vayinstitutes.be/ conclude that this was a ground state phenomenon? PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Judith Klinman: I will talk about methyl transfer ﬁrst. If you use the labelled methyl group to look at the degree of compaction, basically you are looking at

http://www.sol the increase in force constant at the methyl group when it is bound versus free. And you can look at it kinetically, to see what eﬀect mutants have, particularly a

host on on host residue that sits right behind the sulphur that is transferring the methyl group. In our ﬁrst study, we showed that as you decrease the size of the group, the isotope eﬀect approaches unity, which is what you see in solution, and the rate went down 103 fold. We concluded that the transition state — and this was very important

http://www.worldscientific.com/worldscibooks/10.1142/ — was more crowded in the wild-type enzyme, and when we mutated that essential tyrosine to either Phe or Ala, things moved apart and the transition state was more expanded. That would ﬁt with the idea of the transition state being critical. But we went one step further: we simply looked at the binding of the S-AdoMet (S-adenosylmethionine), once again with tritium, three tritiums, in the methyl group to increase sensitivity. A lot of the eﬀect that we had seen in the wild type kinetically was present in the ground state. We did not expect to see that,

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World but it suggests that in the actual binding of this large, extended cofactor (bringing us back to ideas that Jencks developed years ago) the methyl group is presumably NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY forced into a conﬁguration in the ground state where it is quite restricted, either in its bond rotation — carbon-carbon bond rotation — or hyperconjugation. We don’t

have the full physical explanation. But we then performed these extended ab initio calculations with Todd Martinez (Stanford) using his GPU methods. Looking at it a little diﬀerently, he and Heather Kulik (Martinez lab) saw that the converged inter-nuclear distance between the methyl group and the inserted substrate October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 95

Session 2: Discussions 95

depended on a huge number of atoms in the ab initio region. This is why I showed the picture with the scaﬀold together with all the soft squishy stuﬀ in the middle. It is almost as if there are so many binding interactions that, in the ground state

ternary complex you can force things closer than their van der Waals distances. You

must pay the price somewhere: it is likely in the remote binding interactions but you need a lot of atoms to do that. That is a surprising result, but I am not all that confounded because in the tunneling situation you also need to get things very close. But there you do so transiently, by dynamical sampling. That’s the model. Here, you are seeing all this compaction in the computation of the energetic minimum for the ground state ternary complex. So we need to really ask the following question: distribution is allowed. distribution how, with de novo enzymes, can we reproduce the extended interactions that are

going to bring the reactants so close? We know we need a lot of functional groups for catalysis, we need to go one step further, and we need to get the interacting 9180 atoms really close. only. No further No only.

John Gerlt: Can I just follow up, Judith? In other words, did S-AdoMet bind more tightly to the mutant?

Judith Klinman: No, it does not. (Note added in proof : At the time of the confer- vayinstitutes.be/ ence, we had measured Kd values for the mutants using single tryptophan variants PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- of the methyltransferase, seeing a very small trend, with the mutants binding the S-AdoMet slightly less tightly. We have now measured Kd for the binding of S-

http://www.sol AdoMet to the WT methyltransferase and see an overall ca. 10x weaker binding in the direction Kd (Tyr) < Phe < Ala. That is the cofactor begins to bind less

host on on host tightly to the mutants.) The whole idea is that you label the methyl group with three tritiums, and you look at the effect of binding on the discrimination for or against tritium in the bound S-AdoMet with the wild type or the mutant. And you can measure the discrimination between protium and tritium. That tells you what is happening in terms of the force constants at the C-H bonds within the trans- http://www.worldscientific.com/worldscibooks/10.1142/ fered methyl group. And we see that the change in force constant is greater with the wild type. When you make a mutant of a residue behind the sulphur atom, a residue involved in the transfer of the methyl group, the discrepancy between what is happening in solution and in the enzymes starts to disappear. So you actually need that conserved tyrosine pushing on the sulphur. You know there were hints in the literature earlier that there were very strange binding effects in proteins. I remember that Vernon Anderson published some data years ago and everyone said ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World that it must be wrong (and I must confess I was one of those persons). But in this NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY instance, the unusual effect that we are seeing appears to be in the formation of the enzyme complex.

John Gerlt: It is not manifested in the binding constant?

Judith Klinman: The binding constants are not that diﬀerent for S-AdoMet in the mutants. (See note added in proof above.) October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 96

96 Exploring Enzyme Families and Enzyme Catalysis

Kurt W¨uthrich to Judith Klinman: Did you replace this key tyrosine with phenylalanine? And if yes, what happened?

Judith Klinman: Yes, that was one of our mutants, tyrosine to phenylalanine. The tyrosine is probably important because you have a hydroxyl group hydrogen

bonding to a glutamate fairly far away. That keeps the orientation fixed so that the β-methylene can be in the correct position to create this compaction. If we generate phenylalanine, we get some elongation between the methyl donor and acceptor atoms in the computation and diminution in the binding isotope effect. The big difference arises when we put alanine into that position. Now we get an internuclear distribution is allowed. distribution

distance that is not very diﬀerent from what would be seen in solution. There does

not seem to be much compaction left. There is still some with phenylalanine, but not as much as with the constrained tyrosine. Phe is funny because it is of course 9180 much more hydrophobic than tyrosine. only. No further No only. Don Hilvert to Judith Klinman: Along that line, concerning binding, in the phenomenal 500- or 700-fold isotope eﬀect that you see on C-H activation, can you actually rule out the possibility that the labelled molecule, the deuterated molecule, does not bind in a non-productive way? vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Judith Klinman: I cannot tell you anything about the actual binding. The structure that I showed you is the enzyme without the bound substrate. But why would deuterium bind non-productively where protium does not? This is what is so nice http://www.sol about these isotopic labels. All we are changing is the reduced mass of the bond. Honestly, I haven’t thought about that as a possibility because it is not a real host on on host substitution. It is just putting in an isotope. What is so phenomenal is that the isotope eﬀect, which is normally 80 anyway with the wild type enzyme — a very high value — goes into a regime where, in our mind, the donor and acceptor are just too far away. Within a full tunneling model, it is like the deuterium is facing a http://www.worldscientific.com/worldscibooks/10.1142/ wall. The wavelength is just too short to go over that distance and the enzyme has become more rigid and cannot recover the wild-type distance. I would love to have some ideas on how to test that further. We are doing many things like X-rays at room temperature to see if we can look at changes in dynamics and H-D exchange in the double mutant. We have high-pressure eﬀects that suggest it is actually more rigid than the wild type. All these things have to be tested now.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Jason Chin to Don Hilvert: I have a question about the experiment where you do

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY computation and directed evolution. Is there an experiment in the literature where, starting from the same protein scaﬀold, someone has done directed evolution on the parent protein? Has someone done computational design and then directed evolution and compared both processes directly? I have seen a lot of papers where you do the computation and you do directed evolution: the computation clearly does something, the directed evolution does something more. But I would be curious to October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 97

Session 2: Discussions 97

see a comparison starting from the parent. Is that experiment in the literature? So maybe you can comment on both what you think would be the value of such an experiment and whether it has been done.

Don Hilvert: That precise experiment has not been done. David Baker’s lab car-

ried out a control experiment on retroaldolases that required a reactive lysine and Kemp eliminases using a scaﬀold, basically introducing mutations to mimic the composition of the active site that was generated computationally but without any computational information. And indeed, in those libraries he found active enzymes but at a much lower frequency. The activities were very similar, so I predict that if distribution is allowed. distribution

you use these hits as a starting point for evolution, you would get pretty far. For

this type of simple model reaction, that may be a viable approach. If you would like to do something more diﬃcult, my guess is that it will be hopeless because, in our 9180 experience, in screening or selection experiments where you start with no activity,

only. No further No only. you’re in the noise at the beginning and you end up in the noise.

Kurt Wüthrich: to Jason Chin: These modified proteins give unlimited possibilities for the use of NMR spectroscopy, but the production yields must be sufficient. Can you make milligrams of these proteins with single amino acid residue modifications? vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Jason Chin: We can routinely make milligrams of proteins, certainly in E. coli.We have, for example, solved a number of X-ray structures of proteins bearing post- translational modifications — acetylated proteins, ubiquitinylated proteins — using http://www.sol approaches that we have developed. The limitation, at least for us, is not in this area. The limitation is in finding questions well enough molecularly defined that host on on host you know that the outcome is going to be interesting. We could solve hundreds of X-ray structures, as could Tom and other people who have technologies for modifying proteins, but the number in which the questions can be reduced to something molecularly interesting is less clear. For the purposes of NMR, yield is a problem. http://www.worldscientific.com/worldscibooks/10.1142/ You would like to put in things like traceless isotopes. There is, in principle, a conceptual problem because you would like an isotopically labelled amino acid identical to a natural amino acid, but you can’t get a synthetase to recognize an isotopically labelled version of the amino acid. You have to put in a protected version, and deprotect it after the synthesis, so that you end up with the natural version. We have done that in ways that give you a completely native protein having a single isotope label at a single position in the protein. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Kurt W¨uthrich: Do you still get good yields after these additional steps?

Jason Chin: Yes. For example, we have done experiments on a lysine side chain

to which we have attached a para-nitrophenol group via a carbamate. The amino acid goes into the protein and we believe that, during the puriﬁcation process, the nitro group is reduced and a spontaneous fragmentation process yields the lysine back. So we have been able to isotopically label a single lysine in a protein, take October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 98

98 Exploring Enzyme Families and Enzyme Catalysis

the NMR spectrum, and see a single peak, for eﬀectively any protein you want to express in E. coli. For example, you could label an active site lysine and screen for drugs that modify the chemical shift in that one position. This could be achieved

without having to deconvolute, or know the assignments of, the spectrum. It’s not

something we’re done, but if anyone wants to collaborate on that, we can make the protein.

Don Hilvert: Along the same lines, I read a paper recently by Nediljko Budisa in Berlin who claimed that many of these modiﬁed aminoacyl-tRNA synthetases are extremely ineﬃcient compared to the natural enzymes. Competition in vivo with distribution is allowed. distribution

natural processes or normal suppression can be quite severe. Yields of protein are

low and also not very homogeneous. Could you comment?

9180 Jason Chin: There are experiments in the literature where people have done in vitro pyrophosphate exchange, the ﬁrst step of the aminoacylation process in the only. No further No only. synthetase enzyme. Some of these experiments are quite interesting because they suggest that even though the exchange is well characterized, in vivo when you add, let’s say, one millimolar amino acid to the cell and characterize the protein by mass spectrometry — which tells you the identity of the amino acid in the vayinstitutes.be/ protein, and MS-MS which tells the position in the polypeptide where that amino PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- acid is incorporated — those experiments show very clearly that what ends up in the protein is correct and you are able to quantify the yield. So that’s all ﬁne. The

http://www.sol issue is that, in vitro, when you purify the synthetase enzyme and then ask how well do those unnatural amino acids bind to the enzyme active site, in some cases

host on on host that number is quite high. And in some cases, at least in Nediljko Budisa’s hands, the claim is that in that step of the reaction, the natural substrates are competitive. But that does not make this true for the overall catalytic process. Some of those could be dead end complexes and that would be consistent with the observation that what you are selecting on is the overall process and the ability to put the http://www.worldscientific.com/worldscibooks/10.1142/ amino acids into proteins. So it depends on what question you are interested in asking. If you’re asking can we make proteins containing these amino acids and make milligrams per litre, purify them, and then ask questions or do experiments in cells, such as the ones I focused on here today, I think the answer is clearly yes. If you’re interested in what the exact enzymatic properties of the enzymes are, and I point out again that those in vitro measurements are only one step of a multistep process and there’s really no controls that even say that those enzymes are behaving ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World in the way that they are in cells or in fact that they’re folded correctly in vitro, NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY I think that’s interesting characterisation. But I’m not really sure what we learn beyond that.

Ken Houk to John Gerlt: I want to ask more about the pathway docking procedure that you used. It was a quite impressive way to explore the function of an enzyme. The background of my question has to do with the fact that, in general, docking has October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 99

Session 2: Discussions 99

a rather bad name because of its inaccuracy in calculating the binding energies of diﬀerent kinds of molecules. Often, many docking programs will give micromolar to millimolar binding with a whole variety of substrates that might ﬁt in the pocket.

And the second issue is that, when you are looking for function, the enzyme is

evolved to stabilize the transition state, not the substrate. I wonder if you can elaborate on this pathway docking procedure, which proves to be so good?

John Gerlt: Let me try to answer as Matt Jacobson would, since I am here representing a large-scale project. Docking is notorious for being not particularly good at predicting speciﬁc substrates. It was used initially in the pharmaceutical industry to distribution is allowed. distribution

try to identify potential lead candidates. That is not as demanding as, for example,

screening against the KEGG database to ask which metabolites bind best and could be substrates for an enzyme of unknown function. When you dock into a single ac- 9180 tive site, cases in which the best predicted hit is the substrate is not very high. So

only. No further No only. the question is how could you improve that? This might be done by actually docking

to a series of binding sites that constitute a metabolic pathway. We are presently considering catabolic bacterial systems where there are transport systems to bring the catabolite compound in, which is subsequently catabolised. Many transport systems have solute binding proteins in the periplasm and those binding sites can vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS also be used for docking. They also have the advantage that if you produce the

- proteins, you can quantitate binding; you do not have to quantitate catalysis. And this can be done in a relatively easy way by looking for thermostabilisation by which

http://www.sol you can screen a large number of compounds. But the idea is that in fact, if you dock to multiple enzymes in a metabolic sequence, perhaps no docking to any one

host on on host enzyme would give the right answer. But if you look at trends for the docking hits, chemoinformatic methods can analyse those (and it is part of our project to do so), enhancing the ability to decide whether the most likely candidate is a substrate for the metabolic pathway. That is how we produced a trans-hydroxyproline beta¨ıne, the substrate for that racemase. Again, none of this is perfect yet. If it were, the http://www.worldscientific.com/worldscibooks/10.1142/ problem of predicting functions of unknown proteins would be easy. But we are certainly using that docking to two, three, four, ﬁve, six successive active sites with the idea that for the substrates, elements of speciﬁcity will be conserved through the pathway, allowing a better choice of compounds to test. The point of binding versus catalysis is very well taken. Matt Jacobson recently published a rapid report in Biochemistry, I think, about distinguishing between false-positives, which bind

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World but are not catalytically active, and compounds that are real substrates. It was a QM/MM (Quantum Mechanics/Molecular Mechanics) approach sorting out why NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY some compounds predicted to bind were not very good substrates, presumably due to electrostatic eﬀects in the interaction of the various possible substrate candidates

with the active site. I don’t believe this has been totally sorted out. This docking doesn’t work every time but that is why we’re trying to improve it.

Judith Klinman: I just want to get back to the QM issue and how many atoms October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 100

100 Exploring Enzyme Families and Enzyme Catalysis

are going to be needed to really assess what is going on. In the catechol methyltransferase, Todd and Heather needed to extend the number of atoms in the QM region way beyond what is usually done to get a true picture of what is happening

in the binding. I would like to put this out as a major and important new direction

to be tested in other enzymes. This is just one system, yet the theoretician Walter Thiel has been examining another class of enzymes, once again extending the QM region in a stepwise fashion to see how each new area changes ﬁnal conclusions. If we are going to use QM, what are the criteria for concluding whether observations are related to function? Methodology will increasingly include many more atoms than the standard 20. distribution is allowed. distribution

Ken Houk: I am guilty of using quantum mechanics. What you describe is a great problem in that QM, highly accurate for small systems, is combined with 9180 force ﬁeld or MM methods to try to deal with the protein, and in particular to

only. No further No only. include dynamics so that one can calculate not just one but millions of diﬀerent

conformations and average them. While these techniques have been known for some time, they still are not very successful, unfortunately. My lab uses QM to predict the ideal arrangement of functional groups with respect to each other. In fact, for the substrate, precise geometrical arrangements in the transition state are vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS extremely important for catalysis, as mentioned by Judith Klinman. Deviating by

- a few tenths of an A˚ is suﬃcient to lower the rate of catalysis substantially. Todd Martinez is a pioneer at using GPUs for quantum mechanics. Actually, this is

http://www.sol much more common for molecular dynamics. We use GPUs for molecular dynamics too. This takes advantage of the gaming industry’s huge audience for the latest

host on on host machines having very inexpensive processors. This proved to be quite useful for molecular dynamics, but more diﬃcult to apply to QM. We need to apply more accurate methods like QM to larger systems. As this will be more common in the future, one can hope to increase accuracy. Nevertheless, the problems we are discussing are huge, both in terms of the number of atoms involved and molecular http://www.worldscientific.com/worldscibooks/10.1142/ dynamics, namely the necessity to include millions of conformations. The prospect of computation replacing the experimental approach is neither immediate nor even likely in the next decades.

Discussion among all attendees

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Judith Klinman: I would like to use one of my own slides (Fig. 1) in generic fashion to address the question of how enzymes work. Combining experiment and theory NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY is essential at this point. We constantly go back and forth, testing one against the other. I want to use an example on this slide to facilitate discussion. Here, an

internal region of the protein, coloured on the ﬁgure, is surrounded by a scaﬀold which may be pretty rigid. But inside, approximately 600 atoms, a huge number, are interacting with each other. It is the entire ensemble that is essential for catalysis. This instance involves compaction between the methyl group donor and the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 101

Session 2: Discussions 101

distribution is allowed. distribution

9180 only. No further No only.

Fig. 1. We can begin to test structural features capable of generating a short inter-nuclear distance between the methyl donor and acceptor using COMT as a model scaﬀold. Computations by Kulik and Martinez indicate the need for an extended QM region (shown by the atoms in color below),

vayinstitutes.be/ to obtain an energetically converged ground state structure for the ternary complex of COMT. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- acceptor. So now you can start to do in silico experiments. You can take away a single side chain, one at a time, perhaps in combinatorial fashion later, and calculate http://www.sol how close the donor and the acceptor can get. Depending on what is seen, we can then go back to the laboratory to test these out. This kind of interactive prediction host on on host testing, from the computer to the lab and back, is the way we have to go. Another comment concerning compaction: a young man came up to me in the coﬀee break, and he said: if you look at Morse curves, how in the world do you get things so close? We know from physics and physical chemistry that the energy

http://www.worldscientific.com/worldscibooks/10.1142/ to get something shorter than the van der Waals distance is very large, so that’s another question that can be asked. What are all these remote interactions and compensating effects in proteins that can be translated into very specific reactive configurations? I’d like to put that out as a really important question.

Gebhard Schertler: I would like to present two slides very far from this topic but nonetheless related. We work on arrestin, a protein binding to G protein-coupled

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World receptor. We have mutated every amino acid into alanines and tried to stabilize the interaction. We want to make a stronger binder to arrestin. On this graph, the red NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY things bind less well and the blue things bind better. On the next ﬁgure, we include a structure. The better blue binders are very spread out, as are the sub-optimal reds

in the interface. Intuition suggests changing the interface to get diﬀerent binders, but these are too far away. What is the explanation? The blue patches are all far from the binding site. This is similar to what Judith Klinman just said: that things far away inﬂuence what is happening. In this case, it is most likely that we have two October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 102

102 Exploring Enzyme Families and Enzyme Catalysis

states of the protein. Disruption of the blue core with mutants favors an activated state of arrestin: an example of activation from afar. In directed evolution, where do the residues change? Are they changing in the ﬁrst or outer shell? I predict

that a lot of later evolution will happen in the outer shell. When we compare β1 β and 2 adrenergic receptors, the innermost shell contacting the ligand itself is more or less identical. There are no different amino acids. In the outer shell, things are different. There, even small molecules like adrenaline and noradrenaline can bind differently. All these things have to do with the dynamics of the molecule. Events in the outer shell matter because they change protein dynamics. distribution is allowed. distribution

Frances Arnold: You are right. Experiments show that when you evolve an enzyme

for higher activity, allowing mutations to be randomly distributed across the protein, many improvements are found far from the active sites. Though this has been 9180 observed hundreds of times, the exact mechanisms remain to be found. In this

only. No further No only. respect, I think Ken Houk presented some very interesting data today.

Don Hilvert: Following on Frances Arnold’s comment, active site mutations often have greatest impact on activity both favorable and unfavorable. This is to be expected given the starting assumption that interface residues have the largest eﬀect. vayinstitutes.be/ But it is true that proteins are attractive evolutionary objects because you can PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- shape them through distant interactions and transmit that information to a binding pocket. http://www.sol Richard Lerner: In evolving antibody binding, you often start with a pretty poor binder. Under selective pressure, it mutates by somatic random process. When you host on on host go from micromolar to nanomolar dissociation constants, you see both active site mutations and outer-shell mutations. Almost every somatically mutated antibody has outer-shell mutations. Schultz published a rather long paper about this in Science, saying that the general concept is not very complicated and that it is http://www.worldscientific.com/worldscibooks/10.1142/ entropically very good to make the active site a better ﬁt. I do not think that anybody has put it as elegantly as Judith Klinman in terms of compaction but you cannot ﬁnd a somatically mutated antibody that doesn’t make outershell mutations. I have a more general question, not so much for the people in the room but for those who want to get a Gestalt of what we all do. Most people here are chemists including myself. Chemists like to make things and therefore, if there were another similar meeting of people who want to discover things, they might say to us: why do ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World you bother to even make things for the good of mankind because there is so much

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY which remains to be discovered? And the answer is: because we want to learn the principles. But an interesting question would be: how many important molecules, large or small, with say the importance and potency of steroids and prostaglandins,

do the people in this room believe remain to be discovered? Or have we discovered 95% of the important ones? Let’s just deal with small, very potent molecules. What percentage would you believe we have discovered (to Don Hilvert)? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 103

Session 2: Discussions 103

Don Hilvert: People in the pharmaceutical industry always say that their hit rates are much higher if they search natural products libraries than if they start with small molecules from combinatorial libraries. But I do not think they represent

probably more than 50% of the hits.

Richard Lerner: We’ve discovered 50% of the potent molecules, then?

Frances Arnold: When you ask what has been discovered and what remains to be discovered, are you talking about structures or function? Most people care about discovering functions. distribution is allowed. distribution Richard Lerner: I would say functions, Frances, quickly followed by structure. For this purpose, let us just quickly call them functions. 9180 Frances Arnold: So then the context of the question is important: it depends on only. No further No only. what you are asking the system to do.

Richard Lerner: OK. I am asking for function, for example anti-inﬂammatories.

Frances Arnold: But this is a non-natural function because it is something a human vayinstitutes.be/ is interested in but not necessarily the biological system that made it. And the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- same is true for proteins, so many proteins to be discovered that Nature has not made because she could care less. http://www.sol Chris Walsh: With regard to how many natural products remain undiscovered, there are people who do calculations based on the discovery rates of antibiotics of host on on host clinical relevance for human bacterial infections. One in ten Streptomyces make a molecule called streptothricin.The rate of success is 1 in 1000 for streptomycin, 1in104 for erythromycin, and 1 in 106 for daptomycin. And so the calculation has been made to ﬁnd the next important molecule, you might have to screen 1 in http://www.worldscientific.com/worldscibooks/10.1142/ 108. When the pharmaceutical industry stopped screening, they were testing about 3 × 106 molecules per year. So, under classical conditions of screening and looking for new cultures, they were never going to come up with the next new molecule. On the other hand, if you ask how many bacteria are there in the soil, you can count 1018–1019. By that calculation, 98% of the important natural products in terms of antibacterials could still be discovered because we have not screened more than 2% of the soil bacteria that have a history of producing molecules. That’s just one ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World area of therapeutics. What I say is based on an assumption about how rare truly

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY interesting molecules have been, and whether that’s an extrapolatable line.

Richard Lerner: Is somebody going to discover a new steroid in man?

Brian Roth: Yes.

Chris Walsh: If I were looking for interesting new molecules, I would not look in humans. I might look at the microbiome of the gut. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 104

104 Exploring Enzyme Families and Enzyme Catalysis

Brian Roth: My lab does lots of screening and we really love to screen natural products so if anyone here has any, please send them to us, the more interesting the better. And let me say that the approach that we take is to screen hundreds

of molecular targets in a parallel fashion. I think you’re right: if you take the

pharmaceutical approach, which is one target at a time, the hit rate is very low. But if you have a large, genome-wide coverage of a particular target class, particularly the so-called orphans, then every time we do a screen, we ﬁnd really interesting things. Our problem is having the manpower to analyse the results suﬃciently rapidly to disseminate them. distribution is allowed. distribution

Chris Walsh: That is a good and unusual problem to have.

Brian Roth: It is, unfortunately, true. I know there are just so many orphan targets

9180 that, on the basis of knockout data, have really interesting eﬀects. At least in mice, there are really interesting steroid-like molecules that remain to be discovered. I can only. No further No only. expand on that tomorrow. The other point relates to directed molecular evolution. I think we are the only laboratory that really has done this with GPCRs to any great extent. We were evolving a GPCR to bind an unnatural ligand, and be activated by it, to use as a tool eventually. This turned out to be extraordinarily valuable vayinstitutes.be/ for many laboratories. By using saturation mutagenesis, we could screen millions of PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- mutants within a couple of days, and found all of the interesting ones were actually in the binding pocket. So the uninteresting ones that caused non-speciﬁc activation

http://www.sol were outside the binding pocket and had basically allosteric eﬀects. But those that changed the pharmacology of the receptor (which were the ones we are interested

host on on host in) were, as far as we could tell, right in the binding pocket and oriented towards the ligand. The situation may be a little different in other cases, depending on the particular reason for which you are trying to design the protein and whether these outer-sphere mutations that have an entropic effect are going to be effective. But we found that if you are trying to evolve a protein that binds an unnatural ligand, http://www.worldscientific.com/worldscibooks/10.1142/ then, at least in our experience, those were the best.

Richard Lerner: How many enzymes are there in the universe?

Brian Roth: I don’t know.

Richard Lerner: Maybe 104 more? ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Chris Walsh to Judith Klinman: A question for Judith Klinman. I guess I have NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY not thought about compaction very often in terms of catalysis. My question relates to the maturation of molecules like the green and red ﬂuorescent proteins. On the

one hand, triad of residues — serine, tyrosine, glycine in a row — get converted auto-catalytically or by auto-modification to the fluorophore in GFP or red fluorescent protein. But I understand that auto-modification only happens from a folded nascent precursor protein and not from the denatured protein. So is that a case October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 105

Session 2: Discussions 105

of compaction? Presumably not, and a lot of calculations have been done on that auto-modiﬁcation. So tell me how that is diﬀerent.

Judith Klinman: Proteins fold and as they fold, they generate active sites that have a very large number of functional groups, which could also be part of the green

ﬂuorescent protein maturation. Actually, in my view, there is no counterpart in small molecule chemistry because entropically it is simply prevented: you can’t get that many functional groups in one place at one time. So that is step 1. That is how proteins have evolved and of course the energy is paid at the time that the protein molecule is synthesized through GTP hydrolysis. But that is not the end of the distribution is allowed. distribution

story. That is a static ground state structure. And then the question is what else do

we need to get these enormous rate accelerations. Having all the functional groups present does not seem suﬃcient. Now we have to go system by system, but the 9180 model that has emerged — at least in my lab from the hydrogen tunneling studies

only. No further No only. — is that these proteins are actually moving all the time. They are sampling many

diﬀerent ground states, and it is only those ground states that give you all the precise interactions, all the precise hydrogen bonds, electrostatics, etc., that allow catalysis to proceed so quickly. It’s that level of dynamic motion. Of course, we think of proteins now as existing not as a single structure but as a family of conformational vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS sub-states. Any perturbation you make will aﬀect catalysis, whether or not it is

- a remote mutation. There is no question that putting a lot of functional groups around the bound substrate gives you a large rate acceleration. I think you could 8 9 10 http://www.sol probably get a 10 -, 10 -, 10 -fold rate acceleration. I do not think you are ever going to get to a 1025 fold rate acceleration from simply putting these active site

host on on host residues together. There I think what you need is this distribution of conformational sub-states. That being said, in the methyl transfer reaction we appear to be seeing a lot of compaction in the ground state, at least at the ternary complex, but that is just one aspect of catalysis. I do not think that there is any way out of this kind of thinking right now. The problem is how do we show it computationally and http://www.worldscientific.com/worldscibooks/10.1142/ experimentally and bring this all together? That is a challenge for going forward.

John Gerlt: In the complex, what fraction of the protein and substrate molecule is compacted and what fraction is relaxed? You are talking basically about a diﬀerence between the entropy of bringing things together and then, presumably, additional energy — enthalpy — is used to bring about this compacted state.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Judith Klinman: Before we did the methyl transfer experiment, I would have said

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY that it was all in the dynamics, in some fraction of the total protein, and I could not have given you that number. That made it diﬃcult to test the ideas, as well. It’s just a transiently formed sub-state. But in the case of the methyl transfer,

it looks as if there is enough going on in the ground state so that you can get it (the compacted state) as the dominant enzyme form. So then the question arises regarding charge eﬀects: in C-H activation, the charge properties are very diﬀerent October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 106

106 Exploring Enzyme Families and Enzyme Catalysis

from S-AdoMet-dependent reactions where you have the sulphur S+ and nucleophilic substrate. The electrostatics are going to contribute to this phenomenon. I can’t give you a number. All we can do currently is look for changes in the conformational distribution and correlate them with catalytic eﬃciency. This is being done more

and more using biophysical tools.

Marina Rodnina to Judith Klinman: I also have a question for Judith. How common are these tunneling eﬀects in enzymes? In other words, can it be that we are describing enzyme function in an entirely wrong way if we are not taking them into account? distribution is allowed. distribution

Judith Klinman: We have been looking at tunneling for about 25 years. I would

say that there is evidence that virtually every C-H activation has a strong tunneling 9180 component. It is much harder to say that in proton transfers from oxygen to oxygen, since you don’t have the same tools available. But I would say it is very prevalent. only. No further No only.

I think the reason that my group started to think about dynamics is that once you invoke hydrogen tunneling, the movement within the chemical step is just wave function overlap and then all the motions that give rise to the chemistry have to lie within the heavy atoms of the protein. My guess is that this is probably true vayinstitutes.be/ for all catalysis. So yes, there is something wrong with the idea of transition state PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- stabilisation. It’s just looking for all the catalysis in the enhanced binding at the transition state, and not giving us the physical principles that allow these enormous

http://www.sol rate accelerations.

Ray Stevens: A question for Ken Houk. First, an observation: you made a comment host on on host that docking is not very reliable for enzymes, and sometimes we hear kinases have anywhere between 5 and 15% success rates. Whereas with GPCRs, not talking speciﬁcally about enzymes per se, the hit rate is closer to 60 to 80%. You get much better hit rates with docking than with enzymes. And then Richard Lerner made http://www.worldscientific.com/worldscibooks/10.1142/ the observation about enzymes, where you see some mutations on the inside (in catalytic antibody enzymes) but you also ﬁnd a lot of mutations on the outside. Brian Roth made the observation that most of the mutations are occurring directly in the binding site. So I am curious, from a computational perspective: how do you use this knowledge concerning the hit rates and the position of the mutations? Does it have something to do with dynamics? How do you use this and how do you connect those dots? ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Ken Houk: We can’t use that information directly in a computation. A computation is built on the physics of interactions, so we would like to use quantum mechanics since we think that is the right physics, but it is too time-consuming computationally to be used for all systems. There is another complication, which is that the sites of mutation often aren’t directly related to catalysis. You can measure increased catalysis, but, for example, in experiments that I was trying to help explain, Codexis October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 107

Session 2: Discussions 107

is trying to develop a commercially viable catalyst to be used in somewhat unnatural conditions — that is, mixtures of water and DMSO for example. The mutations that occur are contributing to the lifetime of the enzyme and to other factors that are

diﬀerent from catalysis. That makes it diﬃcult to mix in directly with computation.

When we do a computation, we are trying to essentially measure the free energy difference between the bound substrate and the transition state in the enzyme. I didn’t describe any calculations like that today, but that would be the ultimate in using computations to explain catalysis. To do that correctly, unfortunately, it is not possible to use a lot of empirical information. What we need is to have much more accurate force fields if we are going to use classical mechanics or faster distribution is allowed. distribution quantum mechanics. In both fields, progress is constantly being made but we still

have to wait for those developments to occur.

9180 Jan Steyaert: I have a hypothesis rather than a question. If conformational sampling

only. No further No only. is indeed so important for catalysis, I would think that conformational antibodies

that recognize these enzymes would act as allosteric modulators. Is there evidence for that?

Richard Lerner: Jim Wells has published a series of very nice papers where he vayinstitutes.be/ actually selected antibodies that are allosteric modulators. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Jan Steyaert: But those were large conformational changes. This is more subtle, no? http://www.sol

Richard Lerner: 700 A˚2 of protein surface. That is pretty large, yes. host on on host Tom Muir: I have a question for the protein evolutionists. We’ve had some beautiful examples of using in-laboratory evolution to achieve new functions. In one example, Jason Chin is really looking for a synthetase with new speciﬁcity in terms of an

http://www.worldscientific.com/worldscibooks/10.1142/ unnatural amino acid that you can bind to the synthetase and activate. But it is a slightly different situation with a catalyst. You obviously want to have activity toward a specific reaction, but you also would like to have some promiscuity with respect to substrate scope. I wonder what the state of the art is and the current thinking about how one balances these two effects. To me it seems that there are interacting effects: the enzyme is going to evolve toward the particular substrate that you happened to use, not necessarily toward a whole panel of related substrates.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World What is your thinking on that?

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Frances Arnold: This is the sort of question I spend a lot of time thinking about. When chemists say, “I love lipases because they will work on anything,” I answer,

“What they do (hydrolysis) isn’t very interesting, but they do have a broad substrate range and reasonably good selectivities.” The real trade-oﬀ is usually between having a broad substrate scope and having the regioselectivity or enantioselectivity that you really want on each substrate. Right now, it’s just trial and error. It is October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 108

108 Exploring Enzyme Families and Enzyme Catalysis

very hard to know aprioriwhether you’ll have broad scope or not. Another group of people would say that doesn’t matter, because all you need is a quick screen and a couple of mutations to ﬁne-tune the catalyst for any substrate or selectivity that

you want. If you buy that argument, then you really don’t need to have super broad

range, but you have to be able to access the substrate scope that you need.

Don Hilvert: If you want high activity, you have to have very high speciﬁcity as well. Many computational designs are attractive in this regard because you get a foot in the door: you have a promiscuous catalyst that you can then potentially evolve in many diﬀerent directions. distribution is allowed. distribution

Kurt W¨uthrich to John Gerlt: John, you now have this marvellous machinery set

up, and the PDB includes at least 5300 domains of unknown function. How long

9180 will it take to annotate just this sample of structures with your machinery? You should not work in empty space, here you have material to test your machinery. only. No further No only.

John Gerlt: We are developing the machinery to allow such de-orphanization to occur. We spend a lot of time worrying about the proper way to evaluate the success of the Enzyme Function Initiative. Is it evaluated by how many PDB structures

vayinstitutes.be/ we can de-orphanize ourselves, or are we developing strategies and procedures to PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- do that? I would say that we are not necessarily a machine to crank through every structure and de-orphanize it, but what we are trying to do is to develop tools and strategies so that other people can do this. I would say that a problem with many http://www.sol PDB structures is that they were presumably determined in an unliganded state. One of the proteins I talked about today, PDB 2PMQ, was actually crystallized in host on on host what can be called a dockable state. After the fact, the structure was determined with the ligand bound and found to be virtually superimposable on the unliganded structure. I would argue that in cases like that the success rate for docking should be high. And we would not necessarily be interested in all of those. We alone are http://www.worldscientific.com/worldscibooks/10.1142/ not developing a machine that is going to be totally responsible for doing this. I don’t wish to be evasive; our success will be developing pathway docking. Because docking is not a hundred percent accurate in each case, is there a way to do multiple docking experiments and successive catalytic events that enhance that rate? This is not a challenge that one large-scale project is going to solve.

Kurt W¨uthrich: Should I translate this by saying that you are developing a car but

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World don’t want to run the risk of driving it on the highway? NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY John Gerlt: No, not at all, but it is undergoing design continuously.

Chris Walsh: Maybe you have a car but you do not have any gas to put in it.

Wilfred van der Donk to Jason Chin: This question is for Jason Chin. We are doing some incorporation of unnatural amino acids using the methodologies that you and October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 109

Session 2: Discussions 109

others have developed. In our very limited experience thus far, success seems to be very variable. In one position, for the same amino acid, the same tRNA, and the same aminoacyl-tRNA synthetase, incorporation is great. In the same peptide at

another position, we do not seem to get much suppression eﬃciency. So are there

any rules? Is it understood why that is and is it something that you see as well? Are we just doing things wrong?

Jason Chin: There is a series of papers by Miller in JMB, probably in the seventies, that describe codon context eﬀects in amber suppression, for example. We have done things like develop ribosomes that direct more eﬃcient incorporation at amber stop distribution is allowed. distribution

codons. And there are other strategies that address this, which seem to even out

some of those eﬀects that you see if you do a simple experiment with a synthetase and tRNA. The simplest thing to do is to change the codon context, if you can, 9180 around the amber stop codon. There are rules based on this paper by Miller that

only. No further No only. will tell you what to put directly after the amber stop. Of course in some cases that

may result in an amino acid change and in other cases it may just be silent. But that is the simplest thing to do. There may also be contextual eﬀects related to where you are in the polypeptide sequence. You know there is this issue with translational ramping in the N-terminal region of proteins versus further on the polypeptide vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS sequence, as some people have reported. My lab is trying to put together a database

- of our internal experiments and open that up to the community, sharing what the incorporation eﬃciency is at a particular site, which experiments worked well or

http://www.sol which experiments worked not so well, to empirically determine as a community what the rules might be. I think that would be a resource for people. But I think

host on on host the simplest thing for you is to try this codon context.

Henk Lekkerkerker: As a total outsider, I am impressed by these wonderful developments in genome mining, computation, specialist gene lines, etc. In 2007, President Bush predicted that by 2017 the U.S. would be using 35 billion U.S. gallons of http://www.worldscientific.com/worldscibooks/10.1142/ bio-ethanol. It’s generally accepted that only 15 billion of that would come from Midwest corn. That leaves about 20 billion gallons to come from cellulose feedstock. When you talk to people from DuPont, they complain they can get good enzymes (cellulases) only from fungal, not bacterial, sources. They need doses about 25 times as high as for the corn conversion and it is slow. But with these eminent people here, I expect that with the combination of genome mining, calculations, and other approaches, we can solve the problem of DuPont by producing 20 billion gallons of ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World bio-ethanol. We produce about one billion now. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Frances Arnold: It is not the enzymes’ fault. If you compare cellulases from many diﬀerent organisms, they are all pretty much the same in terms of their catalytic

capabilities. There are no huge factors of diﬀerence. So the only opportunity for increasing intrinsic activity is to raise the temperature of the process or to pre-treat the material so that you make the cellulose available. All those things take money. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 110

110 Exploring Enzyme Families and Enzyme Catalysis

The reason not much bio-ethanol is being made is that the price of oil is too low and the price of corn and other feedstocks is too high. Nobody can make money out of it. It is certainly not the enzymes’ fault.

Henk Lekkerkerker: I will tell it to these people, and they will be happy, but I

understand that they are building factories for which the technology is not yet fully developed. Nevertheless, to meet this aim of 20 billion gallons, they are going ahead. I agree that with Shell Gas etc. that the price of oil is low. So it is temperature and not enzymes.

Frances Arnold: Certainly there is some room for improving the enzymes because distribution is allowed. distribution

they are being asked to work under non-natural conditions. But the claims that

you can dramatically increase the intrinsic activity of these enzymes are completely

9180 unfounded. That’s not going to happen. You are going to improve the process context in which they are going to work, and you are going to reduce the cost of only. No further No only. production, which Novozymes and Genencor have been very successful in doing, and they are keeping the rest of the proﬁt in their pocket.

Bernard Henrissat: Building on what Frances has just said, the problem is not so

vayinstitutes.be/ much the enzymes. It is that the substrate has evolved for millions of years to PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- resist enzymatic degradation. You know a tree must survive for an entire lifetime, sometimes hundreds of years, without being destroyed. There has been a lot of invention on the plant side, on the biomass side, to resist degradation. http://www.sol

Frances Arnold: The biomass is very hard to degrade. It has evolved to be hard to host on on host degrade.

Henk Lekkerkerker: I agree it is diﬃcult to make a solution of cellulose and it either has to be highly acidic or highly basic, but the pre-treatment problem seems to be

http://www.worldscientific.com/worldscibooks/10.1142/ solved.

Bernard Henrissat: It is not only cellulose in there.

Henk Lekkerkerker: Agreed.

Lode Wyns: It is a pity that JoAnne Stubbe is not here. She sent in a very nice text, with a lot of interesting manganese bioorganic and inorganic chemistry. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World In going through the text, I noticed that she was the only one who mentioned

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY education and the problem that education is lacking on chemical elements that Nature doesn’t utilize. Looking at myself in the mirror, I say “My own knowledge of those elements is weak.” At one time, we tried to make antibodies against vanadate in order to make kinases, in the same way as they were making antibody enzymes against phosphates to cleave esters. But I experienced that my knowledge, and my inorganic chemistry colleagues’ knowledge, was not suﬃcient to make the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 111

Session 2: Discussions 111

right happen. I guess, when we talk about new enzymes, that there is a whole world of inorganic elements that we don’t deal with. In that respect, education is a major problem. That is what JoAnne wrote. I think that she was the only one to mention

education in this context.

Don Hilvert: That is a very good point, but there are inorganic chemists who are trying to design enzymes. Thomas Ward at the University of Basel, for example, has done very nice work combining the intrinsic reactivity of metal ions with the selectivity that can be oﬀered by protein binding pockets. There’s a lot more to do, obviously: the periodic table is very large. distribution is allowed. distribution

Chi-Huey Wong: Going back to the long distance eﬀects, I have a question for

Judith Klinman and perhaps the others. There are so many examples now of long

9180 distance eﬀects on active site binding and catalysis. So the question is: how far are we from being able to predict the long distance eﬀect on binding and catalysis, and only. No further No only. what do we have to do next to comprehend this?

Judith Klinman: This is the area that makes enzymology a real challenge right now. We can do it on a system-by-system basis. I can give you one example

vayinstitutes.be/ where we have analysed tunneling as a function of a single residue at a dimer PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- interface 25 A˚ away, and can actually show a network that controls the active site compaction. Would I have predicted that in advance? Probably not. The only way we identiﬁed it was by comparing a psychrophile to a thermophile. You know, you http://www.sol have families of homologous enzymes that have to work at very high temperature or low temperature. In that case we got lucky. By comparing the two, we could host on on host actually identify the distal residues in the network. But in general, as I indicated with this the methyl transfer example, we need to go back and forth between in silico computation and experimentation. We don’t have rules yet and I wish I could say more. We don’t even have the required temporal resolution. There has been a http://www.worldscientific.com/worldscibooks/10.1142/ great deal of controversy: enzymes turn over with rates of milliseconds, so for a long time everyone was focussing on millisecond motions to control catalysis. In fact, probably most of the millisecond motions are an accommodation as the enzyme moves through its catalytic trajectory because the enzyme changes its shape as you go from the substrate to an intermediate to the product. Those millisecond motions may just be an accommodation to having diﬀerent structures in the active site. Most of the motions are going to be in that sub-millisecond to nanosecond, ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World even picosecond, regime. It’s tool development that we’ve discussed here. We have

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY to design the appropriate tools to get that temporal resolution. We need spatial and temporal resolution. These are big problems, and I would like to get back to education because I think the way we teach students enzymology is simply not

helping the ﬁeld at all. It is not just bioinorganic or bioorganic chemistry. We really need another approach. Hopefully, the next generation will have the background as well as the tools. We need more people working on the problem. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 112

112 Exploring Enzyme Families and Enzyme Catalysis

Ken Houk: In response to Chi-Huey Wong’s question, the way we are approaching this is not really predictive but it is computational screening. Related to what Ju- dith just said, the only approach so far is to do mutations and then do molecular

dynamics to understand the change in the enzyme average structure, and to deter-

mine whether that improves the active site arrangement or makes it worse. This is unfortunately a very time-consuming procedure, but it is the only way we really can understand how long-distance mutations aﬀect the activity. Unfortunately, sometimes there are loop-motions and so forth that are slow enough so it takes microseconds to milliseconds dynamics to try to account for that. distribution is allowed. distribution

Don Hilvert: How large do the changes have to be to quantify reliably?

Ken Houk: I guess we know anecdotally that a change of several tenths of an A˚

9180 from ideal is suﬃcient to reduce activities substantially. We’re looking to design in smaller diﬀerences than that. only. No further No only.

Frances Arnold: I have been asked over the years by multiple journal editors to speculate on eﬀects of mutations that we discovered by evolution. I always refuse, saying that I want to write a paper of lasting value. That usually works. The reason

vayinstitutes.be/ is that I know that I can tell ten diﬀerent stories and they would all be acceptable PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- but I would not know which one is true. So, now that you are making strides and are answering some of these really diﬃcult questions, I wonder how do you know the story, the explanation that you provide for the eﬀects of a mutation is true? http://www.sol Judith (Klinman), this question is also for you.

host on on host Ken Houk: If I can follow-up and respond to that. I agree that currently it is faster to do random evolution than it is to do computational predictions, but once we are able to do reliable predictions, then we will know that we understand.. I wouldn’t claim we can do that yet. http://www.worldscientific.com/worldscibooks/10.1142/ Judith Klinman: I just want to comment that so many mutations, especially those that are remote from the active site, do not change the X-ray structure at all. That means that you have to look elsewhere for an explanation as to what is going on. When you see structural changes then, I think it becomes really uninterpretable, but that’s where the next round of experiments and computation picks up. When we have profound eﬀects on activity, and the solved structure looks identical, we

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World need to look beyond that to understand what is going on.

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Kurt W¨uthrich: The more this discussion goes on, the more I feel like asking whether we should hope to soon get a general explanation for enzyme mechanisms, or should

we rather aim at investigating the mechanisms of one, two or maybe ﬁve enzymes during the next ten years? What is your opinion? What can we hope for realisti- cally? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 113

Session 2: Discussions 113

Ken Houk: Enzymes take advantage of everything to maximize activity. I don’t think we will get a single answer to that question, other than the very general answer about changing the free energy of activation no matter what the eﬀect. One

can hope that, by studying more systems, we will be able to develop some general

rules about how remote mutations inﬂuence activity, but I do not know what those rules are going to be.

Judith Klinman: I think that there are some fundamental reactions that occur in biology. One of them is phosphoryl transfer, for example, and we have already talked about C-H activation and methyl transfer today. There are some very fundamental distribution is allowed. distribution

classes of reactions essential for life. If I were to choose, I would focus on these very

general classes of bond activation and cleavage, to see if we can in fact develop a core set of principles that will allow us to go forward. 9180 Frances Arnold: But that does relatively little to help the design problem, right? only. No further No only. If you understand the mechanism, it does not necessarily mean that you know how to design it or even how to predict the eﬀects of mutations.

Judith Klinman: Well, I would like to get back to the example of methyl transfer

vayinstitutes.be/ because we deﬁned a core region we think is essential for compaction. Maybe PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- compaction is a design principle. Maybe by going back and forth between in silico and experimental results, we can validate that and see how compaction is generated. And then that is a design principle. http://www.sol

Frances Arnold: But I see that everyone has his pet explanation, and tends to answer host on on host the problem in terms of that explanation without really pursuing alternatives. And, equally, addressing them side-by-side. Probably computation is the best and only way that can be done, but do people really do that? First of all, we also need a tremendous amount of data because prediction is critical, and then we will need to

http://www.worldscientific.com/worldscibooks/10.1142/ predict many times because any individual prediction is an anecdote. Any paper that says that a prediction worked without showing a lot of data could just be showing the prediction that worked on that particular day, while the results of the other nine students whose predictions did not work were not published. We really don’t know what is going on behind the scenes here.

Judith Klinman: I agree with you. I think the ﬁeld has come to be dominated by

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World computation. We need experiments as well. When new ideas emerge, this occurs because you make an observation which can’t be explained by previously existing NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY theories. It is not, let’s throw this out or let’s throw that out. It happens because you have a dominant paradigm and then new data emerge that cannot be explained

by that dominant paradigm. Then you move on to a new way of thinking. I think that often starts from experiments, not from computation, and then you need this interplay between both. Or maybe it can also come from computation. Since I am an experimentalist, I’m obviously biased, but I think it’s the interplay of observations that no longer ﬁt the dominant paradigm. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 114

114 Exploring Enzyme Families and Enzyme Catalysis

John Gerlt: I ask Judith what she regards as the rate-limiting step in bringing together computation and experimentalists. I think that I hear you say that computationalists are not suﬃciently engaged, or is there a lack of interaction with

computationalists? I am not quite sure what the issue is. I don’t mean to be

argumentative. The point I want to make is that this requires collaboration. To bring up again what Kurt (W¨uthrich) said about the Enzyme Function Ini- tiatives being a machine, I could also regard the DARPA project on Computational Enzyme Design (I do not know how it is currently supported) — with design, theoretical analysis, directed evolution, and structure analysis of proteins — as a machine that produced impressive results. I am hoping that our machine will pro- distribution is allowed. distribution duce impressive results. I am in complete agreement that one needs to somehow

ﬁgure out how one can bring together experimentalists and computationalists to analyse and solve this problem. Is the problem achieving this a lack of coordination 9180 or interest? only. No further No only.

Judith Klinman: Part of it is due to a lack of resources. We need the money to go forward. This is a really important area of research. Another part of it, in my experience over the years, has been language diﬃculties. Physical chemists have their own language. These experiments need to be explained to the theoreticians so vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS that they are also fully on board. Therefore I would think that workshops to pro-

- mote that kind of communication could help. But generally the way science works is on a one-on-one basis: you have a question that you cannot answer and then you

http://www.sol look around for someone who has a skill that you do not have, such as computation, and then you move forward. It often takes a long time to lay the groundwork

host on on host so that you fully understand what each person is doing both experimentally and computationally.

John Gerlt: I would say that these machines take some time and thought to put together, but when they have been thought out and put together they in fact can http://www.worldscientific.com/worldscibooks/10.1142/ move forward in a signiﬁcant way. In the United States, there is a “RO1 mentality” for doing science, and bringing together RO1-supported independent scientists to collaborate on a problem is not necessarily the most eﬀective way to solve the problem. Establishing machines, if you can do it, becomes more productive.

Judith Klinman: I think we’re getting into philosophy. There is big science, and there is that individual investigator following his or her nose and seeing where it ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World takes them, but, yes, with resources you would be able to do that. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Gebhard Schertler: There are very good problems for big science. Biologists in the genome project have started to really use that. Looking at new organisms and mechanisms, or ﬁnding functions, is a good opportunity for big science, as long as we can design a screening process. There are also very good reasons for small science and idea-driven science, and I think we need both to work together. Biologists October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 115

Session 2: Discussions 115

certainly have to learn (for example from the physicists) to organize to get the resources, and that often might need a change away from individual contributions that we very much value in biology research.

But let us go back to the idea-driven thing and the interesting discussion. We

talked about time-scale and catalysis and how fast the chemical bond is actually made or broken, and it is quite interesting to think about this. I obviously only know the answer for my pet molecule. And in rhodopsin, this decision is made in 200 femtoseconds. The primary eﬀect that determines the quantum yield in the enzyme is determined in 200 femtoseconds. Now I ask you whether that time scale is not valid for all enzymes. Chemistry operates in a time scale of making and distribution is allowed. distribution breaking bonds, and this must be in a timeframe of femtoseconds to picoseconds

but not longer. You (Judith Klinman) mentioned microseconds and I think this is not correct. But what then happens in biology, and that is what we observe a lot, 9180 is that the chemistry has to be absorbed by the system. Some people have spoken only. No further No only. about a protein quake for example: when you have thrown a ball into an enzyme, you somehow have to dissipate the energy. This is also the case when you put for example a photon in, and I do not know if that happens in catalysis so much. But we also have sub-states, and we have so many intermediate states when we activate

vayinstitutes.be/ rhodopsin because of timing. So at that point chemistry has happened, but the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- protein has to time a process in such a way that it is adequate for a signalling process, which is in this case a protein-protein interaction. So in biology, we deal with very different time-scales. In signalling we need hundreds of microseconds http://www.sol to milliseconds, which allows the diffusion to happen and the partners to find each other. But in catalysis, when you make or break bonds, I think everything is decided host on on host on 200 to 500 femtoseconds, and we need to support the tools, which can address these time-scales.

Judith Klinman: I agree with you completely. The chemistry is happening on

http://www.worldscientific.com/worldscibooks/10.1142/ femtosecond time scales. The motions that I and others are talking about are within the protein. It is the sampling of the diﬀerent conformers that sets up an environment that allows this passage via the activated complex or the trajectory that can lead to product. The actual making and breaking of the bonds is very fast. So the microsecond time scale does not refer to the chemical step, it refers to the class of motions that sets up the environment to allow the bond to be cleaved. You can think of it as an environmental reorganisation; enzymes have evolved as the

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World genius of environmental reorganisation. You use the same thinking for reactions in solution, but what is happening on the enzyme is quite distinct. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Ken Houk: What you say is entirely correct. In transition state theory or in tra-

jectory studies of reactions this happens in a hundred femtoseconds or less. But even for reactions that take hours the actual bond making and breaking still only takes a few hundred femtoseconds. But, as Judith says, it is the probability of that one event happening that can vary, and that probability is inﬂuenced by the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 116

116 Exploring Enzyme Families and Enzyme Catalysis

environment and the dynamic motions that are much slower either in solution or in an enzyme.

Hiroaki Suga: I am always excited by the combination of in silico simulations and in vitro evolution to get catalysis. But right now one needs to start from a certain

protein scaffold. So, how do you pick the protein scaffold and how many proteins scaffolds are available at this moment? How many scaffolds are required to achieve the variety of different catalyses?

Don Hilvert: In the original experiments about 200 scaﬀolds were hand-picked because they were abundant, monomeric and thermostable. Subsequently, it has distribution is allowed. distribution

been possible to really look at every protein in the PDB. I think the success rate

with those other proteins though tends to be much lower, because what you ﬁnd

9180 is that when you introduce ten mutations at the active site of such a scaﬀold you more often than not screw things up. This means that the protein aggregates and only. No further No only. precipitates or can’t be produced. In the Baker lab they have also introduced additional ways of compensating for these destabilization mutations that can then rescue failed designs. With regard to which scaﬀold you choose, the largest fraction of successful enzyme designs has been done in TIM barrels, which is perhaps not vayinstitutes.be/ so surprising since these are very abundant in enzyme families in Nature as well. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- It is very easy to bring catalytic functionality together convergently around the molecule. I guess scaﬀolds like that are going to be the best in the future. http://www.sol Don Hilvert: I think with that we should bring the session to an end and I would like to thank everybody for the vigorous discussion. host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 117

Session 3 Microbiomes and Carbohydrate Chemistry distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 118

distribution is allowed. distribution

Structures determined by the Joint Center of Structure Genomics of

9180 gut microbiome resident bacterial proteins responsible for the degradation of carbohydrates, sulfated sugars, and glycoproteins, only. No further No only. including putative glycosyl hydrolases from Ralstonia eutropha (PDB 3b7f) and Parabacteroides distasonis (PDB 4fj6), Bacillus halodurans glycosidase (PDB 2h56), and Bacteroides thetaiotaomicron sulfatase (PDB 3b5q). vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 119

119

STRUCTURAL BASIS FOR HOST/COMMENSAL-MICROBE INTERACTIONS IN THE HUMAN DISTAL GUT MICROBIOME

IAN A. WILSON Department of Integrative Structural and Computational Biology, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037, USA distribution is allowed. distribution

My view of the present state of human microbiome research 9180 The continued deluge of sequences from genome sequencing eﬀorts are providing only. No further No only.

“omics” and related ﬁelds with unheralded opportunities to explore the richness and diversity of life forms and the fundamental processes that allow organisms to evolve and function in their own particular niches and environments. Metage- nomics has uncovered a vast diversity of microorganisms that colonize the human

vayinstitutes.be/ body, which is estimated to contain over 10 times more microbial cells than human PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

cells. The NIH Human Microbiome Project (HMP) was launched in 2008 with the - goal of exploring these microorganisms and to investigate their role in human health and disease [1–3]. However, microbiome projects to date been have dominated by

http://www.sol eﬀorts to sequence the bacterial genomes and elucidate the types and composition of the bacteria that inhabit speciﬁc environments, such as the human gut. These

host on on host studies have enabled identification of a large number of novel protein families that are specific to the gut microbial communities. Over the past decade or so, the field of structural genomics (SG) was founded to address questions on a genome scale [4–6]. As such, high-throughput structural biology (HTSB) was pioneered to de-

http://www.worldscientific.com/worldscibooks/10.1142/ velop tools and methodologies to tackle biomedical and biological questions on a scale not previously contemplated but critical if the fruits of the genomes eﬀorts were to be realized at the protein structure and function level. These HTSB approaches have indeed enabled broader explorations into the rapidly expanding protein universe, as well as structural coverage of entire organelles, organisms, and collections that inhabit speciﬁc niches, such as the human microbiome. The human body is colonized by vast communities of microorganisms that thrive

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World in speciﬁc niches such as the human gut, skin, oral cavity, and vagina and engage in complex symbioses with the host. These largely commensal microorganisms interact NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY extensively with their host through secreted and membrane-bound proteins and production of small molecule metabolites. However, identifying the precise function

of the proteins involved in the microbiome is a major challenge. As a ﬁrst step in investigating the basis of this complex symbiosis, eﬀorts, such as the HMP, are cataloging and characterizing the diversity of the human microbiota. Substantial genetic diversity has already been discovered in the gut (300 to 1000 unique species, October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 120

120 I. A. Wilson

depending on the methods used) [1, 7] and the activities of their gene products supplement the host-encoded metabolic capabilities and, therefore, are extremely beneﬁcial to the host. Despite the dynamic ﬂux and the species variability between

individuals, certain microbes are consistently detected in high numbers. Diversity is

even greater on the gene level, where organism diversity is compounded by extensive polymorphism in bacterial species. Relative gene composition between species and the distribution of genes within species may thus be more important than species diversity itself [8]. Historically, microbes were typically recognized as infectious agents of disease. However, Koch’s postulates may not hold for complex commensal systems. Human distribution is allowed. distribution health as well as disease is strongly aﬀected by the commensal/host interactions.

While the human microbiome has been implicated in various disorders, only rarely has a single species been identified as a causative agent. In some cases, such as obe- 9180 sity, non-alcoholic fatty liver disease, ulcers and cancer, a specific group of bacteria only. No further No only. within the microbiome may be involved [8, 9]. In other cases, such as inflammatory bowel disease (IBD) and allergies, aberrant inflammatory responses deriving from an imbalance of commensals may contribute to disease (reviewed in [10]). The role of microbes in these complex disorders suggests that many potential disease

vayinstitutes.be/ targets will be identiﬁed from study of these commensal microorganisms. The gut PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- microbiota are, therefore, a largely untapped source of novel and potentially “druggable” targets with the potential to revolutionize therapeutic strategies for many important human diseases and conditions [11]. http://www.sol

host on on host My recent research contributions on the distal gut human microbiome The Joint Center for Structural Genomics (JCSG) which I direct, has established a robust and scalable high-throughput (HT) gene-to-structure pipeline that has

http://www.worldscientific.com/worldscibooks/10.1142/ already delivered over 1450 novel structures to the community, by either X-ray crystallography or NMR, on a wide range of targets from bacteria to human [5]. The JCSG is currently leveraging its HT platform to address challenging new frontiers in structural biology, including exploration of the human microbiome and other biomedically important areas, such as regulation of stem cells, T cells, and human nuclear receptors through interaction with speciﬁc biological partners (see www.JCSG.org). The JCSG is capitalizing on its extensive experience over the past

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 13 years to develop the best strategies to enhance chances of success, particularly in these large-scale projects. Integrative structural/systems biology methods and NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY technologies developed in PSI-2 to analysis of the complete proteome of a single organism, such as our Thermotoga maritima project, and the subsequent modeling

of its central metabolic network, have been invaluable for developing similar applications to the human gut microbiome [4, 12, 13]. These well-tested “engines” of SG are well suited for exploring metagenomic projects in collaboration with specialists in the ﬁeld. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 121

Host/Commensal-Microbe Interactions in the Human Distal Gut Microbiome 121

Since the start of PSI:Biology (July 2010), the JCSG has focused on microbial gut secreted proteins as its main biological theme project. To date ∼ 3000 human microbiome speciﬁc proteins have been processed and screened for potential bioac-

tive activity. To date, over 300 unique protein structures from commensals from the

human gut have been determined. Most of these are gut-secreted proteins that likely interact with the host and cover a variety of biological important host-microbe interactions, such as carbohydrate processing and assimilation, small molecule interactions, and cell signaling. However, as expected from this type of exploratory project, the specific function of most of these targets remains unknown. A significant portion of the genome of these human gut symbionts encodes carbohydrate-processing distribution is allowed. distribution enzymes. Sequencing efforts are greatly outpacing the functional and structural

characterization of these interesting novel classes of enzymes, but progress has been made on the larger families from this class of proteins. 9180 SG eﬀorts further aim to close this gap by enhancing our understanding of only. No further No only. sequence and structure diversity and the inter-relationship between structure- function. Indeed, JCSG structures determined are increasing our knowledge of the human gut microbiome. For instance, the genome of the gram-negative anaerobe dominant symbiont of human distal intestinal tract, Bacteroides thetaiotaomicron

vayinstitutes.be/ (Bt), was sequenced in 2004 and its 4816 genes have a strong representation of PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

> - glycosyl hydrolases (259 or 5% of all genes), as well as 30 recognized proteins with carbohydrate binding modules (CBMs). At the time of sequencing, > 40% of its genes were annotated as “hypothetical proteins”, indicating that these numbers http://www.sol may be underestimated. Indeed, the JCSG has identiﬁed at least 40 additional (putative) glycosyl hydrolases and almost 100 additional CBMs. These enzymes host on on host play a crucial role in the metabolism of polysaccharides and are intimately linked to the host’s metabolism. To date, the JCSG has solved 79 structures from Bt (out of 262), resulting in an additional ∼ 20% of Bt proteins that can now be annotated based on the insights from these structures.

http://www.worldscientific.com/worldscibooks/10.1142/ This new set of structures has helped us identify new (likely) host signaling proteins, new classes of enzymes, new classes of bacterial ﬁmbriae, and novel families of leucine-rich repeat (LRR) proteins with never-before seen arrangements of LRRs. LRRs are typically found in eukaryotic proteins and are usually involved in protein and ligand binding. The best known-human LRR proteins are the innate immunity receptors, TLRs and NLRs. All known eukaryotic LRR proteins adopt a very similar, torus-like shape. The JCSG structural characterization of LRRs stemmed

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World from unexpectedly identiﬁcation of bacterial LRR proteins when exploring human

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY gut microbiome targets of unknown function. Prior to this, very few bacterial LRR proteins were reported and those that were, are predominantly virulence factors, such as Listeria internalin [14]. To date, we have determined 14 bacterial LRR structures, some of which are revealing unexpected novel structural features and variations compared to eukaryotic LRRs that could not have been predicted from the sequence alone. They show an amazing structural diversity, with almost every one displaying novel structural features. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 122

122 I. A. Wilson

In collaboration with Dr. Koji Nakayama (Nagasaki University, Japan) and Dr. Lucy Shapiro (Stanford University School of Medicine), the JCSG is exploring novel types of pilus found in oral pathogens and other gut bacteria. We have

determined structures of 15 novel pilins, which deﬁne a large diverse protein su-

perfamily in the animal gut. Our structures provide strong evidence for a novel mechanism of pilus assembly [15]. Together, this new knowledge is being incorporated into a metabolic network reconstruction, developed in the collaboration with Drs. Andrei Osterman (Sanford- Burnham Medical Research Institute, San Diego, CA), similar to that accomplished with our ground-breaking T. maritima studies that led to the ﬁrst complete struc- distribution is allowed. distribution tural reconstruction of a metabolic network [13]. B. thetaiotaomicron is now our

new model system for structural and function explorations into the human gut microbiome. 9180 Beyond nutrient utilization and immune modulation, the microbiome also inﬂu- only. No further No only. ences the host through its small-molecule metabolites. Our colleagues at the Ge- nomics Institute of the Novartis Research Foundation (San Diego, CA) and others found a number of metabolites in serum and urine derived from human microbiome [16, 17]. For examples, these studies demonstrated that secondary bile acids are

vayinstitutes.be/ modulated by the gut microbiome. Beyond emulsifying dietary fats, bile acids in- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- fluence gene expression through nuclear receptors like FXR, and are implicated in diseases, such as pancreatitis, diabetes, and are the progenitors of gallstones [18]. Bile-modifying pathways were found in only a limited subset of species [19], and http://www.sol the responsible genes (the bai operon of Clostridia) have been identified [20]. The JCSG has recently determined the structure of BaiA [21]. This study represents the host on on host first structure-function characterization of a human gut microbial enzyme involved in secondary bile acid synthesis and provides key insights into the mechanism of catalysis and uncovers structural features that could be utilized to design modulators of secondary bile acid levels in the human body. Another obvious impact of

http://www.worldscientific.com/worldscibooks/10.1142/ microbial metabolism on the host is the bacterial processing of complex carbohydrates. Human gut microbes contribute to host caloric uptake (estimated at 30%) [9] through breakdown of complex carbohydrates into usable sugars or conversion to fatty acids. Gut microbes also metabolize host surface glycans, eﬀectively recycling shed epithelial cells [22–24]. These examples highlight the important role commensals can play in regulating host metabolism. Furthermore, microbial physiology impacts delivery of drugs, were oral delivery can fail due to metabolism by

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World commensals that can inactivate a drug or metabolite compound, or convert it into

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY harmful products, like carcinogens. The microbial enzymatic repertoire can, therefore, inﬂuence circulating metabolites and drugs and, thereby, inﬂuence the host in locations well outside of the gut itself [16].

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 123

Host/Commensal-Microbe Interactions in the Human Distal Gut Microbiome 123

Outlook for future developments on human microbiomes Understanding how commensals have succeeded in occupying a particular niche without causing disease, while closely related strains or microbiome dysfunction

cause disease provides opportunities for generating strategies not only to defeat pathogenic invasion, but also to harness the full potential of the beneficial effects of the microbiome for human health and well being. Mammalian cells detect microbes through pattern recognition receptors, which can bind microbial-specific components. Commensals modulate the host’s innate immune system despite containing molecules, which bind and activate these receptors. The immune system removes pathogens, but there is mounting evidence that the immune response may be just distribution is allowed. distribution as devoted to maintaining beneficial populations [7]. Recognition of commensal versus pathogen is largely defined by interactions between extracellular proteins of

9180 the host and microbe. SG is therefore playing a significant role in understanding the complex symbiosis only. No further No only. of the human microbiome. Many poorly characterized bacterial and host proteins represent potential therapeutic targets. Microbial enzymes producing bioactive metabolites [16] form pathways where structural information is essential in developing specific inhibitors to evaluate in vivo effects on the host. Likewise, extracellular vayinstitutes.be/ microbial proteins function in cell adhesion, immunity and signaling. Beyond struc- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- tures themselves, the SG high-throughput “engines” (gene to structure to function) can be applied in collaborations with biologists to better leverage the impact of these fascinating metagenomics projects. http://www.sol host on on host Acknowledgments Marc Elsliger is gratefully acknowledged for help with preparation of this manuscript, Dennis Wolan for helpful suggestions, and the NIH NIGMS Protein Structure Initiative, PSI:Biology (U54 GM094586) for support to the JCSG. http://www.worldscientific.com/worldscibooks/10.1142/

References 1. P. J. Turnbaugh, R. E. Ley, M. Hamady, C. M. Fraser-Liggett, R. Knight et al., Nature 449, 804 (2007). 2. C. M. Lewis Jr., A. Obreg´on-Tito,R.Y.Tito,M.W.Foster,P.G.Spiceret al.,

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Trends Microbiol. 20, 1 (2012). 3. M. J. Friedrich, JAMA 300, 777 (2008). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 4. S. K. Burley, A. Joachimiak, G. T. Montelione, I. A. Wilson, Structure 16,5 (2008).

5. M. A. Elsliger, A. M. Deacon, A. Godzik,S.A.Lesley,J.Wooleyet al., Acta Crystallogr. Sect. F Struct. Biol. Cryst. Commun. 66, 1137 (2010). 6. T. C. Terwilliger, D. Stuart, S. Yokoyama, Annu. Rev. Biophys. 38, 371 (2009). 7. L. V. Hooper, Nat. Rev. Microbiol. 7, 367 (2009). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 124

124 I. A. Wilson

8. P. J. Turnbaugh, M. Hamady, T. Yatsunenko, B. L. Cantarel, A. Duncan et al., Nature 457, 480 (2009). 9. F. Sommer, F. Backhed, Nat. Rev. Microbiol. 11, 227 (2013).

10. E. A. Grice, H. H. Kong, G. Renaud, A. C. Young, G. G. Bouﬀard et al.,

Genome Res. 18, 1043 (2008). 11. W. Jia, H. Li, L. Zhao, J. K. Nicholson, Nat. Rev. Drug Discov. 7, 123 (2008). 12. S. A. Lesley, P. Kuhn, A. Godzik, A. M. Deacon, I. Mathews et al., Proc. Natl. Acad. Sci. USA 99, 11664 (2002). 13. Y. Zhang, I. Thiele, D. Weekes, Z. Li, L. Jaroszewski et al., Science 325, 1544 (2009). distribution is allowed. distribution 14.W.D.Schubert,C.Urbanke,T.Ziehm,V.Beier,M.P.Machneret al., Cell

111, 825 (2002). 15. Q. Xu, B. Christen, H. J. Chiu, L. Jaroszewski, H. E. Klock et al., Mol. Micro- 9180 biol. 83, 712 (2012). only. No further No only. 16. W. R. Wikoﬀ, A. T. Anfora, J. Liu, P. G. Schultz, S. A. Lesley et al., Proc. Natl. Acad. Sci. USA 106, 3698 (2009). 17. M. Li, B. Wang, M. Zhang, M. Rantalainen, S. Wang et al., Proc. Natl. Acad. Sci. USA 105, 2117 (2008).

vayinstitutes.be/ 18. A. Moschetta, A. L. Bookout, D. J. Mangelsdorf, Nat. Med. 10, 1352 (2004). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 19. P. B. Eckburg, E. M. Bik, C. N. Bernstein, E. Purdom, L. Dethlefsen et al., Science 308, 1635 (2005). 20. J. M. Ridlon, D. J. Kang, P. B. Hylemon, J. Lipid Res. 47, 241 (2006). http://www.sol 21. S. Bhowmik, D. H. Jones, H. P. Chiu, I. H. Park, H. J. Chiu et al., Proteins 82, 216 (2014). host on on host 22. L. Bry, P. G. Falk, T. Midtvedt, J. I. Gordon, Science 273, 1380 (1996). 23. E. C. Martens, H. C. Chiang, J. I. Gordon, Cell Host Microbe 4, 447 (2008). 24. J. L. Sonnenburg, J. Xu, D. D. Leip, C. H. Chen, B. P. Westover et al., Science 307, 1955 (2005). http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 23, 2014 9:20 WSPC Proceedings - 9.75in x 6.5in master/pg 125

125

CARBOHYDRATE CHEMISTRY AND BIOLOGY

CHI-HUEY WONG The Genomics Research Center, Academia Sinica Taipei, 115, Taiwan Department of Chemistry, The Scripps Research Institute La Jolla, CA 92130, USA distribution is allowed. distribution

9180 My view of the present state of research on carbohydrate chemistry only. No further No only.

Carbohydrates on cell surface exist as glycoconjugates and are involved in protein folding and numerous biological recognition events such as diﬀerentiation, development, cell adhesion, pathogen-host interactions, cancer progression and many other

vayinstitutes.be/ intercellular interaction events. However, most of these biological functions are not PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- well understood, mainly due to the lack of tools available for the study, and as a result the pace of the development of carbohydrate-based therapeutics and diagnostics is relatively slow. In the past 10 years, however, many new tools and methods http://www.sol have been developed and used to solve major problems and create new opportunities in carbohydrate chemistry and biology associated with diseases. As a result, host on on host the ﬁeld of glycoscience has attracted a great deal of attention that stimulates more research activities.

My recent research contributions to carbohydrate chemistry http://www.worldscientific.com/worldscibooks/10.1142/ Our interests are to develop new methods and strategies to tackle the problems of carbohydrate-mediated biological recognition and associated diseases. The follow- ings list our contributions to the subject.

1. Synthesis of oligosaccharides on large scales using glycosyltransferases coupled with regeneration of sugar nucleotides in situ. This strategy is considered to be

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World one of the most practical methods available to date for the large-scale synthesis

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY of oligosaccharides, making possible the clinical evaluation and manufacture of this class of biomolecules. The concept was ﬁrst conceived in 1982 and then further improved, as illustrated in the synthesis of sialyl LeX tetrasaccharide and Globo-H hexasaccharide [1–3].

2. Preparation of homogeneous glycoproteins with well-deﬁned glycan structures. Development of such a methodology is important in order to understand the October 23, 2014 9:20 WSPC Proceedings - 9.75in x 6.5in master/pg 126

126 C.-H. Wong

distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol Fig. 1. Identiﬁcation of glycan markers associated with diseases and development of glycan arrays, vaccines and therapeutics. host on on host

eﬀect of glycosylation on glycoprotein structure and function and to develop glycoprotein pharmaceuticals. Ribonuclease B was used as a model to develop methodologies for this purpose [4–6], where the glycan moiety was trimmed down http://www.worldscientific.com/worldscibooks/10.1142/ with endoglycosidases to a homogeneous glycoform followed by addition of new glycans with glycosyltransfer enzymes. Using this strategy, homogeneous glycoproteins with well-deﬁned glycan structures can be prepared from the original mixtures of glycoforms [6]. Alternatively, glycoproteins can be assembled through native chemical ligation or sugar-assisted glycopeptide ligation, followed by elongation of the glycan moiety as described above [6–7]. The methodology

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World was also used to investigate the contribution of each monosaccharide in a glycan

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY to the folding energetics and stability of the glycoprotein. This study has led to the discovery that the monosaccharide (O-GlcNAc or N-GlcNAc) attached to protein contributes most significantly to glycoprotein folding and stability [8–10], and the finding was used in the development of monoglycosylated hemagglutinins as influenza vaccines with broader protection activity, pointing to a new direction of universal vaccine design. October 23, 2014 9:20 WSPC Proceedings - 9.75in x 6.5in master/pg 127

Carbohydrate Chemistry and Biology 127

3. Automated oligosaccharide synthesis. This is a long-standing problem and a major challenge in the ﬁeld. Development of automated methods for oligosaccharide synthesis will facilitate the discovery research in glycoscience, and toward this goal, we reported the ﬁrst automated oligosaccharide synthesis method using the

reactivity-based approach in a one-pot and programmable manner [11, 12]. We continue the eﬀort to design more building blocks with well-deﬁned reactivity to be used for the assembly of various glycans for glycan array development and study of protein-glycan interaction.

4. Development of glycan microarray for analysis of protein-carbohydrate inter- distribution is allowed. distribution action. The idea of glycan array development was conceived in 2002 and our

laboratory is one of the three groups involved in the development [12, 13]. We started with the use of synthetic glycans covalently or non-covalently attached to 9180 microtiter plates, then to glass slides and to the surface of aluminum oxide coated only. No further No only. glass slides [14]. We also developed new methods for determination of dissociation constant [12, 15] and study of multivalency, including heteroligand binding. This work has provided a new platform for the rapid and quantitative binding analysis of protein-carbohydrate interaction, allowing us to understand the

vayinstitutes.be/ speciﬁcity of carbohydrate binding proteins and to dissect their binding energy, PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- including the analysis and detection of inﬂuenza and antibodies that recognize glycans [16, 17].

http://www.sol 5. Identiﬁcation of glycan markers and drug discovery. To follow the process of biological glycosylation and identify glycan markers associated with cancer pro- host on on host gression, we have developed several molecular probes for use to study post- translational glycosylation and identify speciﬁc markers on cancer cells and cancer stem cells [18–20]. This work has led to a better understanding of EGFR dimerization and intracellular signal transduction related to glycosylation [21],

http://www.worldscientific.com/worldscibooks/10.1142/ and a study of fucosyltransferases 8 overexpression related to cancer invasion [22]. In addition, new glycoenzymes or glycan-binding proteins have been iden- tiﬁed from the use of these probes as targets for drug discovery and development of carbohydrate-based vaccines against cancers and infectious diseases [23–25]. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

With more tools available, especially the tools for better synthesis and imaging, the ﬁeld of glycoscience will be further advanced to facilitate our understanding of the biological glycosylation process in disease progression, and hopefully from the understanding, new therapeutic and diagnostic strategies will be developed. October 23, 2014 9:20 WSPC Proceedings - 9.75in x 6.5in master/pg 128

128 C.-H. Wong

References 1. Y. Ichikawa, Y. C. Lin, D. P. Dumas, G.-J. Shen, E. Garcia-Junceda et al., J. Amer. Chem. Soc. 114, 9283 (1992).

2. K. M. Koeller, C.-H. Wong, Nature 409, 232 (2001). 3. T.-I. Tsai, H.-Y. Lee, S.-H. Chang, C.-H. Wang, Y.-C. Tu et al., J. Amer. Chem. Soc. 135, 14831 (2013). 4. K. Witte, P. Sears, R. Martin, C.-H. Wong, J. Amer. Chem. Soc. 119, 2114 (1997). 5. P. Sears, C.-H. Wong, Science 291, 2344 (2001).

distribution is allowed. distribution 6. R. Schmaltz, S. R. Hanson, C.-H. Wong, Chem. Rev. 111, 4259 (2011).

7. A. Brik, Y.-Y. Yang, S. Ficht, C.-H. Wong, J. Amer. Chem. Soc. 128, 5626

(2006). 9180 8. E. E. Simanek, D.-H. Huang, O. Seitz, C.-H. Wong, J. Amer. Chem. Soc. 45,

only. No further No only. 11567 (1998).

9. S. R. Hanson, E. K. Culyba, T. L. Hsu, C.-H. Wong, J. W. Kelly et al., Proc. Natl. Acad. Sci. USA 106, 3131 (2009). 10. W. Chen, S. Enck, J. Price, D. Powers, E. Powers et al., J. Amer. Chem. Soc.

vayinstitutes.be/ 135, 9877 (2013). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS 11. Z. Zhang, I. R. Ollmann, X.-S. Ye, R. Wischnat T. Baasov et al., J. Amer. - Chem. Soc. 121, 734 (1999). 12. C.-H. Hsu, S.-C. Hung, C.-Y. Wu, C.-H. Wong, Angew. Chem. Int. Ed. 50, http://www.sol 11872 (2011). 13. F. Fazio, M. C. Bryan, O. Blix, J. C. Paulson, C.-H. Wong, J. Amer. Chem. host on on host Soc. 124, 14397 (2002). 14. S.-H. Chang, J.-L. Han, S. Y. Tseng, H.-Y. Lee, C.-W. Lin et al., J. Amer. Chem. Soc. 132, 13371 (2010). 15. P. H. Liang, S.-K. Wang, C.-H. Wong, J. Amer. Chem. Soc. 129, 11177 (2007).

http://www.worldscientific.com/worldscibooks/10.1142/ 16. C.-C. Wang, J.-R. Chen, Y.-C. Tseng, C.-H. Hsu, Y.-F. Hung et al., Proc. Natl. Acad. Sci. USA 106, 18137 (2009). 17. H.-Y. Liao, C.-H. Hsu, S.-C. Wang, C.-H. Liang, H.-Y. Yen et al., J. Amer. Chem. Soc. 132, 14849 (2010). 18. M. Sawa, T. L Hsu, T. Ito, M. Sugiyama, S. R. Hanson et al., Proc. Natl. Acad. Sci. USA 103, 12371 (2006). 19. T.-L. Hsu, S. R. Hanson, K. Kishikawa, S.-K. Wang, M. Sawa et al., Proc. Natl. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Acad. Sci. USA 104, 2614 (2007). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 20. C.-S. Tsai, H.-Y. Yen,, M.-I. Lin, T.-I. Tsai, S.-Y. Wang et al., Proc. Natl. Acad. Sci. USA 110, 2466 (2013).

21. Y.-C. Liu, H.-Y. Yen, C.-Y. Chen, C.-H. Chen, P.-F. Cheng et al., Proc. Natl. Acad. Sci. USA 108, 11332 (2011). 22. C.-Y. Chen, Y.-H. Jan, Y.-H. Juan, C.-J. Yang, M.-S. Huang et al., Proc. Natl. Acad. Sci. USA 110, 630 (2013). October 23, 2014 9:20 WSPC Proceedings - 9.75in x 6.5in master/pg 129

Carbohydrate Chemistry and Biology 129

23. K.-C. Chu, C.-T. Ren, C.-P Lu, C.-H. Hsu, T.-H. Sun et al., Angew. Chem. Int. Ed. 50, 9391 (2011). 24.Y.-L.Huang,J.-T.Hung,S.K.C.Cheung,H.-Y.Lee,K.-C.Chuet al., Proc. Natl. Acad. Sci. USA 110, 2517 (2013).

25. H.-Y. Chuang, C.-T. Ren, C.-A. Chao, C.-Y. Wu, S. Shivatare et al., J. Amer. Chem. Soc. 135, 11140 (2013). distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 130

130

CHEMICAL BIOLOGICAL PROTEOMICS OF BACTERIAL PROTEIN FUNCTIONALITIES IN THE HUMAN DISTAL GUT MICROBIOME

DENNIS W. WOLAN Department of Molecular and Experimental Medicine, The Scripps Research Institute

distribution is allowed. distribution 10550 North Torrey Pines Road, La Jolla, CA 92037, USA

9180 My view of the present state of human microbiome research only. No further No only.

The distal gut microbiome represents a vast collection of commensal bacteria that are essential to human metabolism, immune development and homeostasis, epithelial cell angiogenesis, and protection from pathogenic bacteria inﬁltration [1]. Alterations in gut bacteria populations directly govern and dictate the onset of

vayinstitutes.be/ microbiome-related diseases, such as obesity [2], malnutrition [3], inﬂammatory PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

bowel diseases (IBD) (i.e., Crohn’s disease and ulcerative colitis) [4], diabetes [5], - circulatory diseases [6], and colorectal cancer [7]. Recent in vivo manipulations of the bacteria that colonize the distal gut in mouse models have clearly shown

http://www.sol the enormous eﬀect that these bacteria have on the host. For example, genetically compromised IBD mice have chronic intestinal inﬂammation that can be allevi-

host on on host ated by administration of general antibiotics [8]. Conversely, IBD mice raised in a germ-free environment exhibit no inﬂammation-related problems until colonized with microbiota from a healthy donor [9]. These results, and many others, support the development of targeted therapies as innovative approaches to treat gut

http://www.worldscientific.com/worldscibooks/10.1142/ microbiome-related diseases. Great strides in microbiome research have been made over the last decade and are almost entirely attributable to the pioneering development of culture- independent pyrosequencing technologies. These methods generate vast amounts of metagenomic 16S rRNA and deep DNA sequencing information and have provided valuable insights into the ecological diversity and composition of bacteria that colonize the human intestinal tract [10]. Based on the outcome of these studies and

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World the concerted eﬀorts of the Human Microbiome Project consortium, we now know that the taxonomic composition is highly divergent among individuals, populations, NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY and geography and that a universal “core” microbiome of bacterial components is not likely to exist [11]. Furthermore, we have not yet determined what contributing

factors help shape the microbial composition of the distal gut. Strong evidence exists for environmental influences governing the commensal ecology, as differences in diet (e.g., vegetarian vs. high animal protein) have significant dissimilarities in the highly abundant bacterial species [12, 13]. Conversely, host genetics may play October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 131

Bacterial Protein Functionalities in the Human Distal Gut Microbiome 131

a role as conservation among mother-daughter pairs have more similar microbiota compositions than unrelated individuals [14]. The ability to correlate microbiota alterations to external factors and host genetics will greatly improve our ability to

address the causative or casual links between the gut microbiome and diseases.

We need to improve our comprehensive understanding of the role of the human microbiome and we can readily accomplish this task by the application of novel chemical and biological technologies. Considerable limitations in RNA and DNA sequencing analyses restrict the extent to which they can be used to develop a deeper and more fundamental understanding of the biological and biomedical impact that microbiomes have on human health and well being. Despite providing fascinating distribution is allowed. distribution insights into the species diversity and composition of the colonizing bacteria, these

metagenomic studies can only speculate on the bacterial protein composition and its interaction with the host as well as only qualitatively assess variation among 9180 samples without consideration of other resident microbiota organisms, including only. No further No only. fungi, viruses, and single-cell eukaryotes. Furthermore, the application of diﬀerent sets of universal primers results in signiﬁcant variability in the predicted bacterial composition of an individual sample [15]. As such, new methods, chemical biology tools, and interdisciplinary technologies must be developed to further evolve our

vayinstitutes.be/ biological understanding of host:microbiome interactions and to harness and control PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- the enormous impact that enteric bacteria have on human health and disease.

My recent research contributions on the distal gut human http://www.sol microbiome

host on on host We seek to advance the ﬁeld of human distal gut microbiome research from its original basis in genomics into proteomics and quantitative biology through identi- ﬁcation and characterization of bacterial proteins and enzymatic activities that are expressed within healthy and diseased distal gut microbiomes. Unlike metagenomic

http://www.worldscientific.com/worldscibooks/10.1142/ approaches, our chemical biological studies are not focused on population diﬀerences among commensal bacteria species as the root cause of microbiome-associated diseases. We hypothesize that regardless of the microbial composition within the distal gut, conserved bacterial enzymatic and functional activities are altered in the microbiome and spur inﬂammation, tumorigenesis, and other host-driven responses to aberrant microbiota. Importantly, our results over time will also shed light on whether conservation in function is preserved in the face of diversity of the compo-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World sition of the microbiota within the microbiomes of diﬀerent healthy individuals and populations. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Negotiating the complexity of the microbial protein content is an enormous challenge and requires methods that directly and rapidly sample biological systems, yet

permit the systematic analysis and quantitation of individual classes of proteins. To accomplish our goals, we design and use reactive chemical probes, termed activity- based protein probes (ABPPs), that irreversibly label a variety of bacterial protein families and permit the isolation of these tagged proteins from complex proteomes October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 132

132 D. W. Wolan

(Fig. 1). ABPPs are tremendously versatile and contain two essential elements that enable target proteins to be labeled and isolated from complex proteomes: 1) a reactive group that covalently modiﬁes active sites of proteins with shared catalytic

scaﬀolds and 2) an identiﬁcation tag, such as biotin for isolation and/or visual-

ization of labeled proteins (Fig. 1(a)). These types of small molecule probes have been employed with great success in the identification of dysregulated proteins in a variety of human ailments, such as cancerous tumors [16] and parasitic infections [17] (Fig. 1(a)). An advantage for incorporating ABPP into the proteomic analysis of the gut microbiome is the ability to engineer either promiscuity or specificity into the chemical scaffolds for either general mining of the proteome or for explicit distribution is allowed. distribution searches for protein functionality, respectively.

A preliminary goal of ours is to create a universal proteomic methodology that can be applied rapidly and thoroughly to a variety of gut microbiomes. As such, 9180 we have begun to develop our own covalent modifying chloromethyl ketone (CMK) only. No further No only. probes that covalently adhere to reactive serine and cysteine residues located within enzyme active sites (Fig. 1(b)). Our molecules exploit the use of “click” chemistry so that a range of synthetic substructures, including visualization labels (e.g., ﬂu- orophores) or enrichment tags (e.g., biotin) can be introduced after labeling of the

vayinstitutes.be/ microbiome proteome (Fig. 1). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/

Fig. 1. Examples of general ABPP and “click” chemistry (a) and reaction of a chloromethyl ketone probe with surface-exposed cysteine residue (b). ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY This chemical-based enrichment technique is used in conjunction with liquid chromatography coupled to tandem mass spectrometry (LC-MS/MS) for bacterial

protein identiﬁcation and quantitation (Fig. 2). Our development and application of the powerful combination of small molecule probes and LC-MS/MS proteomics allows us to elucidate aberrant enzymatic activities and functionalities associated with diseased microbiomes as well as provide: (1) a novel chemical biological tech- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 133

Bacterial Protein Functionalities in the Human Distal Gut Microbiome 133

nique to complement metagenomic studies; (2) a foundation on which to map the comprehensive host:microbiome proteomic network; (3) a set of valuable chemical tools directed towards understanding the microbiome biology; and (4) an initial list

of potential therapeutic targets to combat or protect against microbiome-related

diseases. distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host

Fig. 2. ABPP proteome labeling, “click” chemistry, and enrichment. http://www.worldscientific.com/worldscibooks/10.1142/ We have recently employed our novel small molecule CMK-based ABPPs in the identification of commensal bacterial proteins from a “normal” human subject. For example, an LC-MS/MS dataset yielded approximately 500 different bacterial proteins that are primarily involved in lipid, carbohydrate, and protein metabolism as assessed by KEGG pathway classification and an additional 15% of the labeled proteins are listed as hypothetical or putative. Importantly, out of the estimated 400

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World bacterial species that are responsible for production of the ABPP-labeled proteins, metagenomic approaches have not reported the majority of these species as members NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY of the commensal gut microbiota. These results strongly support our chemical-based methodologies as a complementary approach to metagenomics. We are currently

expanding our proteomic approaches to establish a baseline enzymatic landscape of proteins among “normal” human gut microbiomes with a variety of ABPP. We will then interrogate microbiomes from patients with a variety of gut-related diseases in order to elucidate compromised bacterial protein functions. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 134

134 D. W. Wolan

Outlook for future developments on human microbiomes One of the most signiﬁcant questions is if altered microbiota drive propagation of host disease phenotypes. Several lines of evidence suggest bacterially produced

metabolites, including short-chain fatty acids directly control the Treg cells of the host’s adaptive immune response [18]. These recent studies are beginning to address the immense potential influence the microbiota have on our health. Moreover, substantial evidence has revealed that manipulation of the microbiome by external intervention, such as antibiotic treatment of IBD, results in ablation of disease. Un- fortunately, such drugs jeopardize the composition of the gut microbial flora and encourage colonization of the now vulnerable intestinal tract by pathogenic bacteria distribution is allowed. distribution [19]. These findings strongly suggest that new chemical and biological approaches are urgently needed to specifically alter microbial enzymatic function and alleviate

9180 host physiological responses to gut bacteria dysbiosis. Most importantly, they illustrate that such therapeutic development will require specific enzymatic targets, as only. No further No only. nondescript off-target effects may result in drastic changes in the population and/or composition of commensal gut flora and have deleterious effects on host symbiosis. In summary, we need to develop and apply novel and innovative complementary methodologies and approaches to provide higher spatial and temporal resolution vayinstitutes.be/ to the many environmental and genetic variables that contribute to understanding PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- of microbiome homeostasis and enhance our ability to manipulate gut bacteria to combat a variety of human ailments. These advancements will only be possible with paradigm shifts in the approaches and techniques applied to the in vitro, cell-based, http://www.sol and animal model studies associated with microbiome research. host on on host Acknowledgment We gratefully acknowledge The Scripps Research Institute for funding.

http://www.worldscientific.com/worldscibooks/10.1142/ References 1. J. I. Gordon, Science 336, 1251 (2012). 2. P. J. Turnbaugh, R. E. Ley, M. A. Mahowald, V. Magrini, E. R. Mardis et al., Nature 444, 1027 (2006). 3.A.L.Kau,P.P.Ahern,N.W.Griﬃn,A.L.Goodman,J.I.Gordon,Nature 474, 327 (2011).

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 4. J. R. Marchesi, E. Holmes, F. Khan, S. Kochhar, P. Scanlan et al., J. Proteome Res. 6, 546 (2007). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 5. A. Giongo, K. A. Gano, D. B. Crabb, N. Mukherjee, L. L. Novelo et al., ISME J. 5, 82 (2011).

6.E.Holmes,R.L.Loo,J.Stamler,M.Bictash,I.K.Yapet al., Nature 453, 396 (2008). 7.J.R.Marchesi,B.E.Dutilh,N.Hall,W.H.Peters,R.Roelofset al., PLoS ONE 6, e20477 (2011). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 135

Bacterial Protein Functionalities in the Human Distal Gut Microbiome 135

8. S. S. Kang, S. M. Bloom, L. A. Norian, M. J. Geske, R. A. Flavell et al., PLoS Med. 5, e41 (2008). 9. W. S. Garrett, G. M. Lord, S. Punit, G. Lugo-Villarino, S. K. Mazmanian et

al., Cell 131, 33 (2007).

10. L. V. Hooper, J.I. Gordon, Science 292, 1115 (2001). 11. C. A. Lozupone, J. I. Stombaugh, J. I. Gordon, J. K. Jansson, R. Knight, Nature 489, 220 (2012). 12. C. De Filippo, D. Cavalieri, M. Di Paola, M. Ramazzotti, J. B. Poullet et al., Proc. Natl. Acad. Sci. USA 107, 14691 (2010). 13. T. Yatsunenko, F. E. Rey, M. J. Manary, I. Trehan, M. G. Dominguez-Bello et distribution is allowed. distribution al., Nature 486, 222 (2012).

14. P. J. Turnbaugh, M. Hamady, T. Yatsunenko, B. L. Cantarel, A. Duncan et al., Nature 457, 480 (2009). 9180 15. D. A. W. Soergel, N. Dey, R. Knight, S. E. Brenner, ISME J. 6, 1440 (2012). only. No further No only. 16. N. Jessani, M. Humphrey, W. H. McDonald, S. Niessen, K. Masuda et al., Proc. Natl. Acad. Sci. USA 101, 13756 (2004). 17. D. C. Greenbaum, A. Baruch, M. Grainger, Z. Bozdech, K. F. Medzihradszky et al., Science 298, 2002 (2002).

vayinstitutes.be/ 18. P. M. Smith, M. R. Howitt, N. Panikov, M. Michaud, C. A. Gallini et al., PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Science 341, 569 (2013). 19. C. Jernberg, S. L¨ofmark, C. Edlund, J. K. Jansson, ISME J. 1, 56 (2007). http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 136

136

AUTOMATED OLIGOSACCHARIDE SYNTHESIS: FROM INSIGHTS INTO FUNDAMENTAL GLYCOBIOLOGY TO VACCINES AND DIAGNOSTICS

PETER H. SEEBERGER Max-Planck Institute for Colloids and Interfaces, Department for Biomolecular Systems, and

distribution is allowed. distribution Freie Universit¨at Berlin, Arnimallee 22, 14195 Berlin, Germany

9180 My view of the present state of research on carbohydrate chemistry only. No further No only.

Access to pure, deﬁned carbohydrates is key to a detailed understanding of the role of glycans in biological systems. Glycans are of particular importance in immunology and the interactions of the microbiome with the human body. Given the

vayinstitutes.be/ fact that complex glycans cover bacterial as well as human cells it is not surpris- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

ing that carbohydrates play an important role for attachment of bacteria but also - in immunomodulation. In the immunological context, glycan have both beneﬁcial as well as detrimental eﬀects and a detailed understanding of structure–function

http://www.sol relationships is needed. While biological studies involving nucleic acids (genetics) and proteins (pro-

host on on host teomics) have made huge strides since the 1970s due to automated sequencing and synthesis methods, glycobiology has suﬀered from the lack of good general methods for both sequencing and synthesis. Matters are more complicated because the term “carbohydrates” really relates to several rather diﬀerent structural classes of

http://www.worldscientific.com/worldscibooks/10.1142/ molecules. For this reason, the synthesis of glycans has also proven greatly more challenging that that of the other two biopolymers. Glycans are banched and each glycosidic linkage constitutes a stereogenic center, thus requiring region — as well as stereocontrol — two aspects that do not enter into the challenge of oligopeptide and oligonucleotide synthesis. All four of the major classes of glycans, glycoproteins, glycolipids, glycosaminoglycans and glycoslyphosphatidylinsotial (GPI) anchors), are glycoconjugates — a combination of an oligosaccharide and another biomolecule,

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World and complicates their synthesis drastically. At the same time isolation of glycoconjugates is also extremely diﬃcult and no ampliﬁcation processes analogous to PCR NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY exist. Carbohydrate chemistry has been practiced for more than one century now and

increasingly more complex molecules have become accessible as chemical techniques have improved. Until today carbohydrate assembly has, for the most part, remained an art as few highly specialized laboratories have taken months and often years to procure a glycan molecule of interest. While the synthetic molecules have con- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 137

Automated Oligosaccharide Synthesis 137

tributed decisively to our current understanding of structure-activity relationships in glycans, the speed (or better lack thereof) in glycan assembly has been a serious obstacle for glycobiology. Glycobiology needs rapid and reliable access to pure and

deﬁned glycans that can be incorporated into tools such as glycan arrays. In the

case of oligonucleotides and peptides automated solid phase synthesis has produced the desired molecules for many years.

My recent research contributions to carbohydrate chemistry Automated Carbohydrate Synthesis: Based on the precedence of DNA and peptide assembly, the Seeberger laboratory developed, by solving a host of chemi- distribution is allowed. distribution cal problems, the ﬁrst automated oligosaccharide synthesizer [1]. After more than ten years of development and a complete overhaul of the synthetic strategy [2], a

9180 commercial synthesizer (Glyconeer 3.0 by GlycoUniverse) is now available to provide access to many complex carbohydrates in days rather than years. only. No further No only. This platform has been used to access the major classes of glycans up to 30-mers [3] including glycosaminoglycans [4], GPI-anchors as well as O-andN-linked glycans. The glycans prepared by the fully automated synthesis system are ready for printing onto surfaces and particles or for conjugation to carrier proteins for vaccine vayinstitutes.be/ development. Using the methods we developed, non-specialists can access deﬁned PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- sugars for biological or medical applications. http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Fig. 1. The automated solid-phase oligosaccharide synthesis platform Glyconeer 3.0 as basis for chemical glycobiology.

Synthetic Tools for Glycobiology: Rapid access to usable quantities of de- ﬁned oligosaccharides has enabled us to create synthetic tools that have been commonplace in genomics and proteomics research. The Seeberger group pioneered October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 138

138 P. H. Seeberger

the use of carbohydrate microarrays and printed the very ﬁrst glycan arrays using robotic printing [5] to determine the ligands for carbohydrate-binding proteins and to deﬁne oligosaccharide vaccine antigens. We screened blood for anti-carbohydrate

antibodies that correlate with disease patterns. In this context we were able to

demonstrate partial resistance to malaria infection based on anti-GPI toxin antibodies [6]. Blood tests for other infectious agents such as parasites threatening pregnant women and for autoimmune diseases are currently under development. Carbohydrate-functionalized, highly ﬂuorescent nanoparticles were prepared and utilized to image carbohydrate-based targeting in vivo [7]. Investigations Into the Role of GPI Anchors in Biology and Medicine. distribution is allowed. distribution The chemical synthesis of (GPI)-anchor glycans was key to the identiﬁcation of the

malaria toxin [8]. A general approach to synthetic GPI anchors provided the basis for detailed biological studies into the role these complex molecules play in biology 9180 [9]. In addition to the ﬁrst synthetic GPI-anchored prion protein we prepared only. No further No only. a host of other GPI glycolipids. Synthetically derived GPIs served to establish the parameters for calculations concerning the structure of GPI [10]. Biophysical studies using grazing incidence X-ray diﬀraction (GIXD) shed light on the structure of GPI-glycolipids in the lipid bilayer of the cell membrane [11]. Investigations into

vayinstitutes.be/ the role of glycolipid signaling in the inﬂammatory cascade, nerve growth and other PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- biologically relevant are currently underway. Deﬁning Glycosaminoglycan-Protein Interactions Responsible for Cel- lular Signaling. Proteoglycans are major components of the extracellular matrix http://www.sol that surround all mammalian cells. Core proteins anchor glycosaminoglycan (GAG) polysaccharides in the outside of the lipid bilayer. GAGs mediate blood coagula- host on on host tion, virus entry, and angiogenesis as well as many other important biological events by binding to proteins such as growth factors, chemokines, and cytokines. To de- ﬁne structure-activity relationships of GAGs, the Seeberger laboratory introduced a modular synthesis of this class of complex carbohydrates [12] that was the basis

http://www.worldscientific.com/worldscibooks/10.1142/ for automated assembly [4]. Microarrays of defined heparin oligosaccharides [13] served as valuable molecular tools to rapidly identify heparin-protein interactions and have been used widely to establish GAG binding motifs of importance in signaling cascades in biological systems. Synthetic Carbohydrate Vaccines. The presence of specific cell-surface polysaccharides on parasites, bacteria, viruses and cancer has been the basis for three marketed vaccines against meningococci, pneumococci and Haemophylis influenza

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World type b (Hib) that are saving millions of lives each year. Since many pathogens can-

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY not be cultured and their cell-surface carbohydrates cannot be isolated in pure form synthetic access to the desired antigens is an attractive alternative. We have utilized synthetic chemistry to access deﬁned oligosaccharides as a basis for vaccine design [14]. Using our synthetic capabilities (vide supra) we have prepared vaccine candidates against several parasitic and bacterial diseases. Vaccine candidates against malaria [8], Group A Streptococcus [15], C. diﬃcile [16], and Neisseria meningi- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 139

Automated Oligosaccharide Synthesis 139

tides [17] as well as Streptococcus pneumoniae are currently at diﬀerent stages of development. Diagnostic Approaches. Monoclonal antibodies raised against synthetic anti-

gens can recognize bacteria very sensitively and speciﬁcally. Antibodies against the

antigen on Bacillus anthracis are now used in a marketed anthrax detection kit [18] while a test for plague based on an antibody is currently under development [19]. Glycan arrays carrying synthetic structures are used to diagnose diﬀerent infectious diseases as well as unwanted immune reactions to drug substances.

Outlook to future developments of research on carbohydrate distribution is allowed. distribution chemistry

Carbohydrate chemistry will be the deciding technology to further progress in the 9180 ﬁeld of glycobiology. In the absence of ampliﬁcation methods analogous to PCR

only. No further No only. access to pure carbohydrates is restricted to synthesis. The products of these syn-

theses will be important standards for glycan sequencing eﬀorts, they will be used to create ever more comprehensive glycan arrays and other tools. Our fundamental understanding in the creation of glycosidic linkages is still far from adequate, even after 100 years of research. A better understanding of mechanisms and factors vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS that inﬂuence the outcome of glycosylations will eventually allow us to predict the

- outcome of glycosylations reliable and not to have to rely on trial and error. In addition to mammalian glycobiology, glycans that play a role in bacterial commu-

http://www.sol nication and for plant biology will come into focus. In addition to communication, access to ever more complex glycans will allow us to ask fundamental questions of

host on on host structure. The role of glycans in plant structure is paramount and by understanding the forces involved in building complex yet stable molecular assemblies will give us a basis to think about creating architectures based on ﬁrst principles using simply glycans.

http://www.worldscientific.com/worldscibooks/10.1142/ The classes of GPI-anchors and GAGs are areas that will beneﬁt from synthetic access. An atomic understanding of GPI-anchored proteins will allow us to understand how such proteins interact in the cell membrane and get involved in signaling process between cells and into cells. The role of GAGs as key mediators of signals via the extracellular matrix is in principle clear. However, what exact sequences or patterns or sequences of GAGs are responsible for interactions with a host of proteins is still unclear. The next few years will see an explosive growth in information

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World regarding glycan-protein interactions that will become available. With an improved understanding of how GAGs and GPIs are involved in signaling mechanisms, new NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY biology and new modes of therapeutic or prevention will become available. Par- ticularly, relatively weak, temporary arrangements of biomolecules in membranes

will be of interest that include carbohydrate-carbohydrate interactions and interactions of GPI-anchored proteins. In time glycans will take their place next to oligonucleotides and polypeptides as key signaling molecules. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 140

140 P. H. Seeberger

Acknowledgments Generous funding of work in my laboratory has been provided by the Max-Planck Society, the ERC (Advanced Grant), the EU (FP7 program), the BMBF and the

K¨orber Foundation.

References 1. O. J. Plante, E. R. Palmacci, P. H. Seeberger, Science 291, 1523 (2001). 2. B. Castagner, L. Kr¨ock, D. Esposito, C.-C. Wang, P. Bindsch¨adler et al., Chem. Sci. 3, 1617 (2012). distribution is allowed. distribution 3. O. Calin, S. Eller, P. H. Seeberger, Angew. Chem. Int. Ed. 52, 5862 (2013).

4. S. Eller, M. Collot, J. Yin, H.-S. Hahm, P. H. Seeberger, Angew. Chem. Int.

9180 Ed. 52, 5858 (2013). 5.D.M.Ratner,E.W.Adams,J.Su, B. R. O’Keefe, M. Mrksich et al., Chem- only. No further No only. BioChem. 5, 379 (2004). 6. F. Kamena, M. Tamborrini, X. Liu, Y.-U. Kwon, F. Thompson et al., Nat. Chem. Biol. 4, 238 (2008). 7. R. Kikkeri, B. Lepenies, A. Adibekian, P. Laurino, P. H. Seeberger, J. Amer.

vayinstitutes.be/ Chem. Soc. 131, 2110 (2009). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 8. L. Schoﬁeld, M. C. Hewitt, K. Evans, M. A. Siomos, P. H. Seeberger, Nature 418, 785 (2002). 9. Y.-H. Tsai, X. Liu, P. H. Seeberger, Angew. Chem. Int. Ed. 51, 11438 (2012). http://www.sol 10. M. Santer, M. Wehle, I. Vilotijevic, R. Lipowsky, P. H. Seeberger et al., J. Amer. Chem. Soc. 134, 18964 (2012). host on on host 11. C. Stefaniu, I. Vilotijevic, D. Var´on Silva, M. Santer, G. Brezesinski et al., Angew. Chem. Int. Ed. 51, 12874 (2012). 12. H. A. Orgueira, A. Bartolozzi, P. H. Schell, R. Litjens, E. R. Palmacci et al., Chem. Eur. J. 9, 140 (2003).

http://www.worldscientific.com/worldscibooks/10.1142/ 13. J. L. de Paz, C. Noti, P. H. Seeberger, J. Amer. Chem. Soc. 128, 2766 (2006). 14. P. H. Seeberger, D. B. Werz, Nature 446, 1046 (2007). 15. A. Kabanova, I. Margarit, F. Berti, M. R. Romano, G. Grandi et al., Vaccine 29, 104 (2010). 16. C. E. Martin, F. Broecker, M. A. Oberli, J. Komor, J. Mattner et al., J. Amer. Chem. Soc. 135, 9713 (2013). 17. Y. Yang, S. Oishi, C. E. Martin, P. H. Seeberger, J. Amer. Chem. Soc. 135, ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 6262 (2013).

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 18. M. Tamborrini, D. B. Werz, J. Frey, G. Pluschke, P. H. Seeberger, Angew. Chem. Int. Ed. 45, 6581 (2006). 19. C. Anish, X. Guo, A. Wahlbrink, P. H. Seeberger, Angew. Chem. Int. Ed. 52, 9524 (2013). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 141

141

CARBOHYDRATE-ACTIVE ENZYMES IN MICROBIOMES

BERNARD HENRISSAT Centre National de la Recherche Scientiﬁque and Aix-Marseille University Campus de Luminy, 13288 Marseille, France distribution is allowed. distribution My view of the present state of research on microbiomes and

carbohydrate chemistry 9180 Despite their apparently unexciting chemical composition, carbohydrates can form only. No further No only.

an astronomical number of combinations through the stereochemical diversity of the hydroxyl groups that they carry, through the multiple possibilities to assemble monosaccharides to each other, and through the many non-carbohydrate sub- stituents that can decorate the resulting oligo- and polysaccharides. Laine has 12

vayinstitutes.be/ calculated that there exists more than 10 possible isomers for a reducing hex- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

asaccharide [1]. Although not all possible isomers are found in Nature, Nature - has long exploited the stereochemical and physical diversity of carbohydrates for a large variety of biological roles, from carbon reserve (starch, glycogen, laminarin,

http://www.sol etc.), to structure (cellulose, hemicelluloses, pectins, peptidoglycan, chitin, man- nans, glycosaminoglycans, carrageenans, agarose, porphyrans, fucans, etc.), or as

host on on host the mediators of intra- and intercellular recognition within one organism or between organisms [2]. The complex glycan structures borne by glycoproteins or the multitude of bacterial exopolysaccharides mediate the recognition between a host and commensals, symbionts or pathogens. The complexity of carbohydrate recognition

http://www.worldscientific.com/worldscibooks/10.1142/ reaches probably its maximum in the distal intestine where hundreds of diﬀerent bacterial strains digest the colossal diversity of our food and epithelial glycans, while harboring a multitude of exopolysaccharides that participate to the maturation of the immune system [3]. Nature’s utilization of carbohydrates seem only limited by the ability of enzymes to assemble and disassemble a glycan without interfering with another. The number of protein folds being limited, Nature has created carbohydrate-active enzymes

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World (CAZymes) from a small number of primordial scaﬀolds. This creative process has taken place all along the evolution of living organisms and has enabled the emer- NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY gence of a multitude of complex glycans which became substrates for breakdown enzymes. For instance, cellulose was a relatively minor macromolecule when life was

essentially marine. The successful emergence of terrestrial plants was made possible by the exceptional mechanical properties of cellulose. From then on, cellulose became the most abundant macromolecule of photosynthetic origin on Earth, thereby creating an immense new ecological niche for organisms that would be able to di- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 142

142 B. Henrissat

gest plant cell walls. Various microbes and fungi jumped on the new opportunities oﬀered by land plant biomass by “recycling”anumberofpre-existingenzymesthat progressively adapted to the new substrate. This process took place independently

several times from diﬀerent protein scaﬀolds, explaining why nowadays cellulases

are found in more than 10 different families. The evolutionary processes by which novel substrate specificities are acquired from a common ancestor leave traces that can be detected in the sequence of contemporary proteins [4, 5]. Thus, unexpectedly, what is the usual drawback of carbohydrates (their chemical resemblance) is at the origin of their success in the post- genomic era: whilst the fine specificity of DNAses, RNAses, proteases and esterases distribution is allowed. distribution are practically impossible to derive from their sole sequences, the hierarchical classi-

fication system that we have proposed for CAZymes [6–8] allows, already in several cases, the prediction of the carbohydrate substrate based on the assignment to a 9180 family [9]. In principle, this provides the opportunity to examine the glycobiological only. No further No only. profile of an organism based on its genome sequence. However, the acquisition of novel specificities from older CAZymes often groups together in the same family enzymes that act on different substrates and this results in significant problems during the functional annotation of a number of CAZyme-related genes. Today

vayinstitutes.be/ our knowledge of functional variations within CAZyme families is still insuﬃcient PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- to accurately and reliably predict the substrate specificity of all CAZymes identified during genomic or metagenomic sequencing. The organization of the sequence information in clans, families and subfamilies can serve as a guide to select puta- http://www.sol tive CAZymes for functional studies by identifying the poorly explored regions of each family, i.e., where homology-based specificity predictions are unreliable. Our host on on host classification of CAZymes therefore can contribute simultaneously to the rational identification of targets for functional studies and to the utilization of this information to decipher the Glycobiology organisms based on genomic or metagenomic sequence. This virtuous circle will progressively contribute to the emergence of

http://www.worldscientific.com/worldscibooks/10.1142/ Glycogenomics, and will be instrumental to establish a Systems Biology that will capture the peculiarities of carbohydrates.

My recent research contributions to microbiomes and carbohydrate chemistry The spectacular progress of modern DNA sequencing technologies together with the

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World launch of large multinational eﬀorts allowed (i) to isolate and sequence the genome of a multitude of gut bacteria and (ii) to generate metagenomic data of increasing NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY quality and depth [10–12]. During the last 5 years, I have explored the potential of CAZymes in the func-

tional proﬁling of genomic and metagenomic data derived from the digestive microbiota of humans and animals. The main results based on CAZyme family assignment and the resulting broad substrate categorization have led to several observations. First, an early metagenomic study of cow rumen has revealed that rumen October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 143

Carbohydrate-Active Enzymes in Microbiomes 143

CAZymes appear to contain a number of appended carbohydrate-binding modules much smaller than CAZymes from environmental bacteria [13]. This is attributable to the conﬁnement and highly substrate concentration found in the digestive

ecosystems.

Second, we noted that the CAZyme profile of obese individuals was less diverse than that of control subjects [14]. Next, through the analysis of fecal metagenomic data derived from 33 mammalian species and 18 humans, we showed that diet shapes the CAZyme profile in animals and humans [15]. Finally, we recently analyzed the enormous genomic and metagenomic datasets generated by the HMP consortium [12], which sampled not only gut, but also bucal, urogenital, nasal and skin body distribution is allowed. distribution sites [9]. Examination of 493 bacterial reference genomes isolated from different

body sites, showed that carbohydrate degradation capabilities appear more similar within a bacterial taxonomical family than among bacteria colonizing the same 9180 habitat. Yet, the analysis of 520 metagenomic samples from the five major body only. No further No only. sites showed that even when the community composition varies the CAZyme profiles are very similar within a body site, suggesting that the observed functional profiles have adapted to the local carbohydrate composition [9]. All in all, and despite the current limitations due to their incompletely stud-

vayinstitutes.be/ ied sequence-to-function relationships, CAZyme proﬁles appears to correlate with PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- both taxonomy and habitat (lifestyle) of individual organisms (genomes) or communities thereof (metagenomes). Irrespective of the expectable improvements of sequencing technologies, future progress will depend on our capacity to predict with http://www.sol increased accuracy the fine substrate specificity of CAZymes identified in genomes and metagenomes. host on on host

Outlook to future developments of research on microbiomes and carbohydrate chemistry

http://www.worldscientific.com/worldscibooks/10.1142/ In principle, the clan-family-subfamily classification of CAZyme sequences has the potential to determine the glycobiological profile of an organism simply based on its genome sequence. However, the multiple specificities that are often grouped together in a given sequence-based CAZyme family is an obstacle to the simple and straightforward functional annotation of CAZyme-related genes. Whilst this problem can be overcome by the definition of smaller subfamilies [16, 17], our knowledge of functional variations within CAZyme families and subfamilies is still insufficient

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World to accurately and reliably predict the substrate speciﬁcity of all CAZymes identi- ﬁed during genomic or metagenomic sequencing. The organization of the sequence NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY information in clans, families and subfamilies should serve as a guide to select and study candidate CAZymes for functional studies by identifying the poorly explored

regions of each family, i.e., where homology-based specificity predictions are unreliable. Coupled to the discovery of novel families by the glycobiological community, the hierarchical classification of CAZymes therefore can contribute simultaneously to the rational identification of targets for functional studies and to the utilization October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 144

144 B. Henrissat

of this information to decipher the Glycobiology of organisms based on genomic or metagenomic sequence. This virtuous circle will progressively improve and the precise speciﬁcity of most CAZymes that interact closely with their substrate will be

predictable. If coordinated, this eﬀort will contribute to the emergence of Glycoge-

nomics, and will be instrumental to establish a Systems Biology able to capture the peculiarities of carbohydrates. Applications could be massive as carbohydrate polymers constitute abundant renewable materials for use by a variety of industries and for the production of transportation biofuels. Complex carbohydrates also constitute a major fraction of human and animal diet and could be used to manipulate, stabilize or restore distribution is allowed. distribution the composition of the microbial communities along the digestive tract. Due to its

wide variety and limited overlap with the human metabolic pathways, the human microbiome constitutes a potential source of therapeutic drug targets [18]. The 9180 breakdown or modiﬁcation of orally administrated drugs by the microbiota also only. No further No only. attracts attention from a health perspective [19]. Although the microbiota has been linked to a number of diseases, it is presently unclear whether the changes in the microbiota composition are the cause or the result of the disease [20]. The interplay between the microbiota and heath is still in its infancy, but certainly constitutes a

vayinstitutes.be/ mine of opportunities for the future. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Acknowledgments Current research in my laboratory is funded by the following institutions: Centre http://www.sol National de la Recherche Scientiﬁque (CNRS), European Research Council (ERC), European Commission (EC), Agence Nationale de la Recherche Scientiﬁque (ANR), host on on host Fondation Infectiopole Sud, and University of Aix-Marseille.

References

http://www.worldscientific.com/worldscibooks/10.1142/ 1. R. A. Laine, Glycobiology 4, 759 (1994). 2. A. Varki, N. Sharon, Essentials of Glycobiology 2nd edition, 1 (2009). 3. A. El Kaoutari, F. Armougom, J. I. Gordon, D. Raoult, B. Henrissat, Nat. Rev. Microbiol. 11, 497 (2013). 4. C. Chothia, J. Gough, C. Vogel, S. A. Teichmann, Science 300, 1701 (2003). 5. S. Kumar, K. Tamura, M. Nei, Brief Bioinform. 5, 150 (2004). 6. B. Henrissat, Biochem. J. 280, 309 (1991).

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 7. B. Henrissat, G. Davies, Curr. Opin. Struct. Biol. 7, 637 (1997). 8. B. L. Cantarel, P. M. Coutinho, C. Rancurel, T. Bernard, V. Lombard, B. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Henrissat, Nucl. Acids Res. 37, D233 (2009). 9. B. L. Cantarel, V. Lombard, B. Henrissat, PLoS One 7, e28742 (2012).

10. P. J. Turnbaugh, R. E. Ley, M. Hamady, C. M. Fraser-Liggett, R. Knight, J. I. Gordon, Nature 449, 804 (2007). 11. J. Qin, R. Li, J. Raes, M. Arumugam, K. S. Burgdorf et al., Nature 464,59 (2010). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 145

Carbohydrate-Active Enzymes in Microbiomes 145

12. The Human Microbiome Project Consortium, Nature 486, 207 (2012). 13. J. M. Brulc, D. A. Antonopoulos, M. E. Miller, M. K. Wilson, A. C. Yannarell et al., Proc. Natl. Acad. Sci. USA 106, 1948 (2009).

14. P. J. Turnbaugh, M. Hamady, T. Yatsunenko, B. L. Cantarel, A. Duncan

et al., Nature 457, 480 (2009). 15. B. D. Muegge, J. Kuczynski, D. Knights, J. C. Clemente, A. Gonz´alez et al., Sci- ence 332, 970 (2011). 16. M. R. Stam, E. G. Danchin, C. Rancurel, P. M. Coutinho, B. Henrissat, Protein Eng. Des. Sel. 19, 555 (2006). 17. H. Aspeborg, P. M. Coutinho, Y. Wang, H. Brumer 3rd, B. Henrissat, BMC distribution is allowed. distribution Evol. Biol. 12, 186 (2012).

18. B. D. Wallace, M. R. Redinbo, Curr. Opin. Chem. Biol. 17, 379 (2013). 19. I. D. Wilson, J. K. Nicholson, Curr. Pharm. Des. 15, 1519 (2009). 9180 20. L. Zhao, Nat. Rev. Microbiol. 11, 639 (2013). only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 146

146

THE MICROBIOME(S): MICROBIOTA, FAMILIES, FUNCTIONS

ADAM GODZIK Program in Bioinformatics and Systems Biology, Sanford-Burnham Medical Research Institute 10901 N. Torrey Pines Rd, La Jolla, CA 92037, USA distribution is allowed. distribution My view of the present state of research on microbiomes and

carbohydrate chemistry 9180 Complex microbial communities thrive in every environmental niche on Earth and only. No further No only.

recent breakthroughs in DNA sequencing provided us with unprecedented level of information about these communities, eﬀectively rediscovering them both for science and popular imagination [1]. For over a century, microbiology was dominated by a “pure culture paradigm”,

vayinstitutes.be/ where the ability to grow a microbe in a pure culture was seen as a prerequisite to PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

any scientiﬁc study of its function and/or characteristics. While very successful in - identifying microbial causes of many diseases, in cases where individual members are diﬃcult or even impossible to study in isolation it slowed down and sometimes

http://www.sol reversed studies of microbial communities. Existence of such communities, for instance multi-bacterial bioﬁlms, was well known since the invention of the microscope

host on on host — already in 1677 Anton van Leeuwenhoek reported seeing multiple “Animals in the scurf of a mans Teeth”, a claim that was widely questions and almost led to his expulsion from the Royal Society of Lodon [2]. However, only technical advances in DNA sequencing that allowed direct sequencing of environmental samples (metage-

http://www.worldscientific.com/worldscibooks/10.1142/ nomics) brought these observations to the forefront of modern science. Surveys of ocean [3] and human microbiomes [4] made metagenomics and microbiomes a subject of popular press [5]. Especially a realization that humans, and to a varying extend all multicellular organisms, are in fact super-organisms that depend on their microbial components for normal functioning captured public imagination. Human microbiome consists in fact of hundreds of smaller microbiomes, as every surface and even smallest region of our body hosts separate, ofter very diﬀerent microbiomes

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World [6]. An often repeated estimate suggest that in human body, microbes outnumber human cells by a 10:1 margin, and microbial genes outnumber human ones at least NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 100:1. When we include viruses, which outnumber microbes by a similar margin, we are indeed outnumbered in our own body.

Previous microbe-human interaction paradigm viewed microbes as agents of disease and lack of them as a symbol of health, it seems that the opposite is closer to the truth. For instance a ﬁrst symptom of many diseases is diminishing the diversity and numeric count of our microbiome [7] and the absence of some bacterial October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 147

The Microbiome(s): Microbiota, Families, Functions 147

species was found to be correlated to diseases [8]. The most obvious role of human microbiome, especially the gut microbiome, is providing humans with the ability to digest food particles that are normally not accessible to human enzymes. Human

gut bacteria, such as Bacteroides, developed extensive networks of carbohydrate

metabolism able to process complex polysaccharides of plant origin, as well of pep- toglycans (protein-carbohydrate polymers) of animal origin. For rats, 30% of an caloric intake is provided by byproducts of microbial metabolism [9]. The second role is protecting the host, mostly by overcrowding the pathogenic bacteria, but also by priming host immune system. Microbiome also may play other, at this point little understood roles, such as contributing to the maturity of host nervous system distribution is allowed. distribution [10]. Research on human microbiome is still in a very early phase, but it is clear

that our microbes form dense networks of interactions with the host, us, on every level — from metabolites, to proteins to cell-cell interactions. 9180 As much as human microbiome captured the headlines, as microbiomes go it only. No further No only. is not particularly complex or diverse. Many environmental microbiomes display astonishing diversity with for instance at least 9 distinct phyla are usually abundant in soil microbiomes [11], compared to two major ones in the human gut microbiome. Bacterial communities are found in most exotic environments, from the Atacama

vayinstitutes.be/ desert [12], to inside of glacial lakes [13], isolated for thousands or in some cases PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- for millions on years. Novel carbohydrate metabolism enzymes found in human gut microbiome are just a tip of the iceberg of fascinating discoveries we can expect from the barely touched or still to be explored regions of protein universe in other http://www.sol environments, driven by unusual and often unknown chemistry and we can expect many fascinating discoveries from the barely touched or still to be explored regions host on on host of protein universe.

My recent research contributions to microbiomes and carbohydrate chemistry http://www.worldscientific.com/worldscibooks/10.1142/ In my group we have been collecting and analyzing gene repertoires from different microbiomes and developing a family fingerprint approach to directly compare them for the functional content, an alternative to the phylogenetic profiling such as done by the 16S rRNA. By comparing presence/absence of protein families in different functional groups, we can for instance see an functional complementarity between family repertoires

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World in the human genome and in the human gut microbiome, where families lost in the human lineage since the last eukaryotic common ancestor (LECA) can be found NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY in the human gut bacteria, thus making the human+microbiome superorganism having a similar metabolic capacity as LECA (Fig. 1) [15].

We also put a big effort into defining new protein families, specific to given types of microbiomes [16]. Since most protein families defined in public databases were identified from model, broadly distributed microbes, they rather obviously focus on broadly distributed families, which members have universally conserved October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 148

148 A. Godzik

functions. In well studied microbes, over 80% of all proteins can be classified into known families. At the same time, only 40–60% of predicted proteins coded by genes identified in metagenomic studies can be identified, suggesting that microbes

forming microbiomes in many environments could have developed novel functions,

tuned to the specific challenges of their environments. Therefore databases of protein families have to be expanded to include novel families found in newly surveyed environments. We have defined and deposited to Pfam over 400 protein families specific to the human gut environment and we are now using them to analyze and compare microbiomes associated with diseases and study geographical diversity of human microbiomes. distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ Fig. 1. Comparison of bacterial composition of diﬀerent microbiomes, here between human mi- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- crobiomes associated with two forms of inﬂammatory bowel disease (Crohn’s disease and ulcerative colitis): (left) principal component analysis of species distribution [14], (right) relative frequen- cies of proteins from speciﬁc families, here proteins involved in butyrate synthesis (Godzik lab,

http://www.sol unpublished). host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Fig. 2. Human gut microbiome complements the domains lost in the human lineage since the

last eukaryotic common ancestor (LECA). Reconstruction of ancestral domain repertoires suggests that LECA had an extensive repertoire of metabolic functions (black bars), many of which were lost in the branch leading to vertebrates, including humans (red bars). While individual human gut microbes (green and yellow bars) have limited repertoires of such domains, together with the host they almost achieve the same numbers as those in LECA. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 149

The Microbiome(s): Microbiota, Families, Functions 149

Outlook to future developments of research on microbiomes and carbohydrate chemistry We have just begun to survey the world of microbiomes and the amazing diversity

of genes they contain. Even if a relatively small percent of these new genes code for enzymes, we still face billions of novel enzymes, performing potentially novel functions.

Acknowledgment Research described here was funded by the NIH grant U54 GM094586 (JCSG). distribution is allowed. distribution

References 9180 1. National Research Council (U.S.). Committee on Metagenomics: Challenges only. No further No only.

and Functional Applications. and National Academies Press (U.S.), The new science of metagenomics : revealing the secrets of our microbial planet. 2007, Washington, DC: National Academies Press. xii, p. 158. 2. F. N. Egerton, J. History of Biol. 1, 1 (1968).

vayinstitutes.be/ 3. J. C. Venter, K. Remington, J. F. Heidelberg, A. L. Halpern, D. Rusch et al., PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

Science 304, 66 (2004). - 4. S. R. Gill, M. Pop, R. T. Deboy, P. B. Eckburg, P. J. Turnbaugh et al., Science 312, 1355 (2006).

http://www.sol 5. C. Zimmer, New York Times 2010, (2010). 6. P. J. Turnbaugh, R. E. Ley, M. Hamady, C. M. Fraser-Liggett, R. Knight et al.,

host on on host Nature 449, 804 (2007). 7. D. C. Baumgart, W. J. Sandborn, Lancet 380, 1590 (2012). 8. H. Sokol, B. Pigneur, L. Watterlot, O. Lakhdari, L. G. Berm´udez-Humar´an, et al., Proc. Natl. Acad. Sci. USA 105, 16731 (2008).

http://www.worldscientific.com/worldscibooks/10.1142/ 9. D. L. Sewell, B. S. Wostmann, C. Gairola, M. I. Aleem, Amer. J. Physiol. 228, 526 (1975). 10. C. L. Maynard, C. O. Elson, R. D. Hatton, C. T. Weaver, Nature 489, 231 (2012). 11. V. Torsvik, L. Ovreas, Curr. Opin. Microbiol. 5, 240 (2002). 12. A. de los R´ıos, S. Valea, C. Ascaso, A. Davila, J. Kastovsky et al., Int. Microbiol. 13, 79 (2010).

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 13. Y. M. Shtarkman, Z. A. Ko¸cer, R. Edgar, R. S. Veerapaneni, T. D’Elia et al., 8 NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY PLoS One , e67221 (2013). 14. J. Qin, R. Li, J. Raes, M. Arumugam, K. S. Burgdorf et al., Nature 464,59 (2010).

15. C. M. Zmasek, A. Godzik, Genome Biol. 12, R4 (2011). 16. K. Ellrott, L. Jaroszewski, W. Li, J. C. Wooley, A. Godzik et al., PLoS Comput. Biol. 6, e1000798 (2010). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 150

150

N-LINKED PROTEIN GLYCOSYLATION

MARKUS AEBI Institute of Microbiology, Department of Biology, ETH Z¨urich, CH-8093 Z¨urich, Switzerland distribution is allowed. distribution My view of the present state of research on microbiomes and

carbohydrate chemistry: Protein glycosylation

9180 For a long time, carbohydrates as protein modifications have received little attention and very little was known about their functions. This lack of knowledge only. No further No only. was mainly due to the lack of analytical tools that specifically detect and identify complex carbohydrates on proteins. In addition, complex carbohydrates are the product of biosynthetic pathways in which the structure of the product is determined by the specificity of the enzymes (glycosyltransferases) involved and not by vayinstitutes.be/ a linear template such as RNA in the case of proteins. A good example of the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- fact that carbohydrate modifications of proteins have been overlooked is the relatively recent discovery of the dynamic modification of a large number of cytoplasmic and nuclear proteins by N-acetylglucosamine on serine and threonine residues http://www.sol (O-GlcNAcylation) [1]. This protein modification is as frequent and as dynamic as

host on on host protein phosphorylation but the lack of a simple and specific detection system has slowed down the functional analysis of this highly conserved process in eukaryotic cells. Similarly, protein glycosylation was for a long time considered to be eukaryote- specific but research in the last ten years revealed that such modifications are very frequent in prokaryotes and that protein glycosylation is highly divers [2]. Novel an- http://www.worldscientific.com/worldscibooks/10.1142/ alytical tools such as MS-based methods and oligosaccharide-specific antisera have been instrumental in the identification of such novel protein glycosylation processes in prokaryotes. Reverse genetics in model organisms is an excellent tool to study the function of proteins in vivo and this tool can also be applied to reveal the function of carbohydrate modifications. However, this requires a detailed analysis of the biosynthetic processes and the characterization of the different glycosyltransferases that are in- ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World volved generation of protein-linked oligosaccharides. Therefore, genetic tools have NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY only been applied recently in glycobiology but they revealed a large functional diversity of protein glycosylations. Powered by genetic studies of glycosylation processes in model organisms such as yeast and mouse, a large number of human congential

disorders have been recently linked to deﬁciencies in protein glycosylation pathways [3]. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 151

N-Linked Protein Glycosylation 151

My recent research contributions to microbiomes and carbohydrate chemistry: Protein glycosylation The research of my group has focused on the process of N-linked protein glycosyla-

tion. Using the model organism Saccharomyces cerevisiae we have characterized the N-glycosylation pathway in the Endoplasmic Reticulum and characterized many of the steps in the assembly and the transfer of the N-linked glycan to protein [4]. We also studied a homologous process of N-linked protein glycosylation found in the bacterium Campylobacter jejuni and characterized it at a molecular level after functional transfer of the pathway into Escherichia coli [5]. In collaboration with the group of Kaspar Locher, ETH Z¨urich, the crystal structure of the bacterial distribution is allowed. distribution oligosaccharyltransferase (OST) was solved [6]. This structure revealed the molecular basis for the recognition of the conserved N-X-S/T substrate motif and was the

9180 basis for the “twisted amide” model of the reaction mechanism. It is interesting to note that N-linked protein glycosylation became the only. No further No only. most frequent protein modification in eukaryotes: more than 20,000 different N- glycosylation sites are proposed to be glycosylated in the eukaryotic proteome [7]. A short recognition element (N-X-S/T) is a prerequisite for the observed evolutionary trends in N-glycan multiplicity in specific sets of proteins. Due to the fact that vayinstitutes.be/ OST requires a flexible peptide for substrate recognition, the execution of N-linked PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- protein glycosylation before the folding process of substrate proteins is a specific feature of eukaryotic N-linked protein glycosylation as compared to its prokaryotic counterparts. Additional subunits of the eukaryotic OST can slow down the folding http://www.sol process of substrate proteins [8]. In particular, Ost3/6p exhibit oxidoreductase activity and bind specific polypeptides, both non-covalently and via transient disulfide host on on host bonds and thereby maintain the substrate peptide in a “glyosylatable” stage. This shift from the prokaryotic order of events (translocation of the polypeptide – folding – glycosylation) to the eukaryotic order (translocation – glycosylation – folding) results in an increase of potential glycosylation sites. At the same time, the transfer http://www.worldscientific.com/worldscibooks/10.1142/ of the highly hydrophilic oligosaccharide to a polypeptide alters its folding parameters significantly: N-glycosylation extends the folding landscape of polypeptides. However, the folding of polypeptides is not only affected by the altered biophysical properties due to attached oligosaccharide, the oligosaccharide itself is used as an anchoring point for the assembly of a glycan dependent chaperone and folding machinery, the calnexin/calreticulin complex [9]. In addition, the processing of the N-linked glycan is used to signal the folding status of the protein and defined car- ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World bohydrate structures are essential signal for the degradation of not properly folded

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY proteins in the ER of eukaryotic cells [10]: N-linked protein glycosylation is a central factor in the folding of secretory and membrane proteins in the eukaryotic cell. Once in place on properly processed proteins, the N-linked glycans can serve

additional functions and the eukaryotic cell has devoted a whole organelle to the remodeling of protein-bound glycans: the Golgi apparatus. In a species- and cell- type speciﬁc process, N-linked glycans are trimmed and reconstructed by a series of October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 152

152 M. Aebi

diﬀerent hydrolases and glycosyltransferase resulting in the well-known structural diversity of N-linked glycans. Using a novel, MS-based analytical method we were able to quantify site-speciﬁc glycan structures on proteins with multiple N-linked

glycans (C-W. Lin, R. Gauss, S. Fleurkens, I. and M. Aebi, unpublished results),

Our results revealed a site-specific processing of the N-glycan by the Golgi machinery, suggesting a kinetically controlled process that is influenced by the interaction of the oligosaccharide with the covalently bound polypeptide. It is the processing machinery in the Golgi, the structure of the glycoprotein and the location of the attachment site that determines the site-specific oligosaccharide structure. The site-specific structures of N-linked glycans can provide novel functionalities to N- distribution is allowed. distribution glycoproteins.

9180 Outlook to future developments of research on microbiomes and carbohydrate chemistry: Protein glycosylation only. No further No only.

The structural complexity of carbohydrate modiﬁcations will remain a challenge for analytical approaches. New MS-based approaches that are fast and that give a detailed structural information are urgently needed. It will be of outmost importance that the analysis of glycan functions does not remain a domain of specialized glyco- vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS biologists and glycochemists. Once glyco-analytics can be performed in any decently

- equipped laboratory, we will be able to identify a very high diversity of functional roles of protein-bound glycans. It is evident that glycans are the main components

http://www.sol of the extracellular matrix in all cells, pro- and eukaryotic. Cells interact with their environment via their glycan interphase and we will be able to decipher the un-

host on on host derlaying “glycocode”. The biophysical properties of carbohydrates, diﬀerent from those of nucleic acids and polypeptides, are the basis for this “code” but it is evident that carbohydrate-binding proteins, lectins, are the main interpreters. A large number of lectins are encoded in all genomes that are sequenced so far, but for only a few of these proteins we know their biological function. http://www.worldscientific.com/worldscibooks/10.1142/

Acknowledgments Work in my lab has been generously supported by the Swiss National Science Foun- dation, the EU (framework 7) and ETH Z¨urich.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World References

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 1. G. W. Hart, C. Slawson, G. Ramirez-Correa, O. Lagerlof, Annu. Rev. Biochem. 80, 825 (2011). 2. H. Nothaft, C. M. Szymanski, Nat. Rev. Microbiol. 8, 765 (2010).

3. H. H. Freeze, V. Sharma, Semin. Cell Dev. Biol. 21, 655 (2010). 4. P. Burda, M. Aebi, Biochim. Biophys. Acta 1426, 239 (1999). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 153

N-Linked Protein Glycosylation 153

5. M. Wacker, D. Linton, P. G. Hitchen, M. Nita-Lazar, S. M. Haslam et al., Sci- ence 298, 1790 (2002). 6.C.Lizak,S.Gerber,S.Numao,M.Aebi,K.P.Locher,Nature 474, 350 (2011).

7. D. F. Zielinska, F. Gnad, J. R. Wisniewski, M. Mann, Cell 141, 897 (2010).

8. E. Mohorko, R. Glockshuber, M. Aebi, J. Inherit Metab. Dis. 34, 869 (2011). 9. A. Helenius, M. Aebi, Annu. Rev. Biochem. 73, 1019 (2004). 10. R. Gauss, K. Kanehara, P. Carvalho, D. T. Ng, M. Aebi, Mol. Cell. 42, 782 (2011). distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 154

154

SESSION 3: MICROBIOMES AND CARBOHYDRATE CHEMISTRY

CHAIR: IAN A. WILSON AUDITORS:N.CALLEWAERT1,R.LORIS2 (1) Unit for Medical Biotechnology, Inﬂammation Research Center, VIB and Department of Biochemistry and Microbiology, Ghent University, Technologiepark 927, B-9052 Gent-Zwijnaarde, Belgium distribution is allowed. distribution (2) Structural Biology Brussels, Vrije Universiteit Brussel and Department of Structural Biology, VIB, Pleinlaan 2, B-1050 Brussels, Belgium

9180 only. No further No only.

Discussion among panel members Karen Nelson could not make it but was kindly replaced by Adam Godzik.

Ian Wilson: Let me ﬁrst start by asking about the microbiome and this problem of

vayinstitutes.be/ associating it with disease and are we ever going to get to that situation where we PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- will be able to do that?

Adam Godzik: I am sure we are. I think that there are a lot of interesting correla-

http://www.sol tions. What we are still having problem with is whether they are really causes of these diseases or just bystanders. There are some very interesting results on mice

host on on host and model systems, which suggest that this is the driving force, but this is still not proven.

Ian Wilson: What about the sample variation and the difficulty of actually assess- ing. . . this is a real problem, isn’t it? http://www.worldscientific.com/worldscibooks/10.1142/ Adam Godzik: Yes, the diversity is an enormous problem but there are some signatures which survive these differences. There are specific combinations of species and specific combinations of pathways, which are overrepresented or missing despite all the variations.

Dennis Wolan: There are plenty of examples in the literature, particularly of the

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World butyrate producing species, not even species but phyla, that are involved in disease

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY states, lack thereof actually. Jeremy Nichols I think has been doing quite a lot of work, particularly in colorectal cancers, to show that particular metabolisms, like cresol (and cresol sulfate) are overproduced in colorectal cancer patients. So taking these kind of molecules and potentially bating for the proteins that do interact with them is a potential way of going after drug discovery targets.

Adam Godzik: Let me add something. One of the interesting observations people make is that very often disease states are associated with upsets of certain groups of October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 155

Session 3: Discussions 155

proteins of pathways. So very often it seems that they are not causing the disease but, if they are missing, this propagates the disease state.

Bernard Henrissat: I would like to complete this (or try to contribute to this) by saying also, our view of what is going on is limited by the fact of what we sample.

We sample what is coming out. We should be sampling, using animals for instance, what is going on all along the digestive tract. Because sampling of what is going on in the mid colon or at the beginning of this colon is very diﬃcult, and it is rarely done on these cohorts of people.

Dennis Wolan: The resolution that we are at right now is quite minimal. We do distribution is allowed. distribution

know that the diﬀerent types of bacterial species that reside for instance in the

mucosal layer are highly diﬀerent from those in the luminal content. The other

9180 issue is as far as animals are concerned; there is some debate as to how useful they are, particularly as the mucin layer composition is quite diﬀerent in mouse as only. No further No only. compared to human.

Bernard Henrissat: I very much agree with this when I mentioned animals, I was not thinking about mice. We just take a whole animal, such as little piglet or pig.

vayinstitutes.be/ And you can perform some experiments with an animal than you cannot perform PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- in a human person.

Ian Wilson: Anyone else want to contribute to this topic? http://www.sol Richard Lerner: What about mycoplasma? Bacteria live in a tube, mycoplasma

host on on host live on and in the cells. There doesn’t seem to be much attention on mycoplasma.

Adam Godzik: As I mentioned, in one of the unexplored areas, that all the issues about intracellular bacteria are just beginning to be appreciated. We just obtained very interesting results on a possible role of intracellular bacteria in colorectal can- http://www.worldscientific.com/worldscibooks/10.1142/ cer and it definitely seems tumors are infected by specific strains of intracellular bacteria. We don’t know if they are cause or effect of colorectal cancer, but definitely mycoplasma, and many other bacteria of this type, are driving things. They are just not studied yet.

Judith Klinman: I was wondering if bacterial metabolites can serve a function in signaling in mammalian cells. Is anything known about that? (Like bacterial ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World metabolites being used by mammalian enzymes) NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Dennis Wolan: There are some examples, for instance in butyrate-producing Clostridium bacteria, butyrate does block NFκB activity. There are clear examples where metabolites that are produced by bacteria do have an eﬀect on the physiological aspect of the human. A lot of it has to do with inﬂammation, as far as what is in the literature at least. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 156

156 Microbiomes and Carbohydrate Chemistry

Ian Wilson: How are we going to get at all these diﬀerent functions? Do you think we are going to get it through these chemical probes, through structures, or through bioinformatics approaches? How are we going to ﬁgure out all of the diversity of all

functions that are going on in the microbiome?

Bernard Henrissat: I can take a try on this one. I think we need to integrate everything, we cannot leave something aside and say “This is not going to work out”. We will need high-throughput assays for functions, we will need three-dimensional structures, we will need everything to acquire enough knowledge so we can go into prediction. We cannot leave anything aside — we will take it all. distribution is allowed. distribution

Dennis Wolan: Along those lines, the Human Microbiome Project is expanding into

trying to incorporate a wide variety of “-omics” kind of technologies: metabolomics,

9180 transcriptomics, genomics, proteomics, etc. to decipher the kind of differences that are observed in disease versus normal states. However, there is definitely something only. No further No only. to be said about taking examples of particular proteins or working from a bottom- up kind of approach to understand the functional significance and how they can umbrella into other functions of bacterial proteins.

vayinstitutes.be/ Adam Godzik: I agree that this problem is very complicated. There are now two PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- broad areas/ways to approach it. One uses larger studies, where we hope that by comparing large cohorts of disease versus healthy people, you will notice the single biggest signal. Then identify the signal and feed it back into model systems. There http://www.sol are several studies like this on inflammation and several other diseases, where they start from observations in large human cohorts and then go to mouse models and host on on host try to see whether single identified features can be used to modulate the system. On the other hand, there are a lot of ad hoc random trials and with some of them, like probiotics for instance which are not regulated, something can be discovered but it also generates a lot of noise. http://www.worldscientific.com/worldscibooks/10.1142/ Ian Wilson: Bernard said something which I thought is quite provocative: that function is not determined by the fold. What do you exactly mean by that? You are probably talking about fine specificities because there are carbohydrates processing enzymes, signature sequences. . . Can you explain a little more?

Bernard Henrissat: I will take just one example that is been under our nose for so

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World many years, that is lysozyme. If you did not know what α-lactalbumin is doing, you would probable predict it to be a lysozyme. By fold, by sequence similarity, by NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY everything. It is only when you inspect the catalytic machinery that is not there that you may have a hint that it is no longer functioning as a lysozyme. But you

cannot guess what the new function is. You could not guess that one ancestral fold has been has been recruited as the binding partner for a glycosyl transferase to alter its acceptor/donor speciﬁcity. This is what I mean there: the fold will not predict the function necessarily. Often it will do. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 157

Session 3: Discussions 157

Ian Wilson: I get that. But I was really referring to the carbohydrate-processing enzymes. Are there so many diﬀerences between the folds? So the fold is already telling you it is a carbohydrate-processing enzyme. What you don’t know is prob-

ably the substrate speciﬁcity. Is that what you are referring to?

Bernard Henrissat: It is right. It is “determining the speciﬁcity”: knowing that you cleave a sugar is not suﬃcient. You would like to know which sugar. You would like to know whether you have an ensemble of enzymes, in a operon or a cluster, that target a complicated substrate, or the substrates that gets in the microbiota. It is very complex and needs the co-operation of many enzymes. Knowing that you distribution is allowed. distribution

have a sugar-cleaving enzyme is not fully satisfying. I would like to have more!

Ian Wilson: Are there any questions from the discussion panel members? 9180 John Gerlt: I have a question about the carbohydrate enzymes. Are these only. No further No only. carbohydrate-active hydrolases and transferases excreted and used to degrade polysaccharides so that we can use these as carbon sources? What is the diversity of catabolic pathways within the human microbiome species because you only talked about the hydrolases, but there are kinases, aldolases, dehydrogenases, for

vayinstitutes.be/ example. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Bernard Henrissat: I can only speak about the enzymes I know. There is a great variety also of glycosyl transferases that are elaborated by the bacteria of our mi- http://www.sol crobiome, for instance. The reason for this variety is presently unclear but when we look at the number of diﬀerent glycosyl transferases that are encoded in the micro- host on on host biome, there is an enormous diversity of carbohydrate structures that are produced at the surface of the bacteria in our gut. One of the ideas I have on the role of such a variety is for protection against phages. If all our bacteria in the gut had the same carbohydrate structure at their surface, they would probably be very sensitive to

http://www.worldscientific.com/worldscibooks/10.1142/ phage attack — what you have in a dairy factory where you have a single species of Lactobacillus for instance. By having a variety of carbohydrate structures you ensure resilience of the microbiota.

John Gerlt: But I am so curious about the catabolism of carbohydrates by the gut bacteria. Is there much known about it? Is it a simple catabolism to maintain the human microbiome? Is it producing the hydrolases so that we then can utilize a

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World wider variety of dietary oligo- and polysaccharides?

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Bernard Henrissat: There is a little bit of this. It is estimated that about 10% of the calories that the microbiome is deriving from the food we eat is utilized by the

(human) host. It is only 10%. But I believe that of the main “raisons d’ˆetre” of the microbiome is simply that we have a lot of carbohydrates, because we eat a diet that is complex and we just derive very little ourselves. When it (food) gets in the distal gut, there is temperature, water, heat and microbes can ﬂourish. So I think October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 158

158 Microbiomes and Carbohydrate Chemistry

that the main “raison d’ˆetre” of the microbiota is just for itself more than for the host. There has been a good level/amount of co-evolution (we accept this or this bacteria because they do not harm us), there are some very good things, but the

(bacterial) ﬂora is there mostly for itself.

John Gerlt: So they could be producing signaling molecules or other molecules to interact or so that we can interact with them, so as to receive some positive beneﬁts, that is not just a carbon source.

Gebhard Schertler: Can I ask a question here? You said we have realized that many bacteria form ﬁlms. Are carbohydrates actually a substrate to sustain ﬁlms distribution is allowed. distribution

of bacteria and make them more speciﬁc because we want to retain certain bacteria

and not others. Maybe that is a way to look at it? 9180 Markus Aebi: No the carbohydrate is the matrix that creates the bioﬁlms mediated only. No further No only. by lectin-carbohydrate interactions, and even carbohydrate-carbohydrate speciﬁc interactions.

vayinstitutes.be/ Discussion among all attendees PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Ian Wilson: We now can open this up to the general discussion for any of the speakers. Maybe I will start with asking Chi-Huey (Wong) about the eﬀect of http://www.sol the glycosylation. You talked about stabilization but that was for a particular sequence, where you had to have a phenylalanine upstream. Do you think it is host on on host general or for speciﬁc cases? And how can we use this information to think of stabilizing glycoproteins in general?

Chi-Huey Wong: I think phenylalanine is of small signiﬁcance, but in some cases

http://www.worldscientific.com/worldscibooks/10.1142/ you see glycosylation will aﬀect folding, but not of every protein. Hemagglutinin is one of them. There is no phenylalanine there, but Asn27 is essential for folding. If there is no glycan there, it is not enough. The metal transfer I mentioned, the antibody, none of them has the aromatics but they still aﬀect the folding pathway.

Ian Wilson: But you said for the hemagglutinin that there was a certain residue for which glycosylation is required for folding. We have bacterially expressed hemagglutinin and it seems to fold just ﬁne. So how do you reconcile that? ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Chi-Huey Wong: I didn’t see that. Did you express the same sequence?

Ian Wilson: Yes, with the bacterially expressed form of the hemagglutinin. We haven’t published that yet.

Chi-Huey Wong: We expressed the hemagglutinin in human cells with a mutation at Asn27 but couldn’t get it. We use for example PNGase to remove the glycan, we October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 159

Session 3: Discussions 159

also see the structure fall apart. That one (Asn27) is essential for folding and there is another one, Asn142, that is close to the binding site for the sugar (sialic acid). If you don’t have the glycan there, you miss out on the stereospeciﬁcity because

hemagglutinin will recognize both the α(2-6) and α(2-3) linked sialic acids. These

are the two glycosylation sites that we identiﬁed. The other four are not really essential for stabilization but I think they are involved in function.

Dario Neri: I have a question for Ian Wilson that may be also relevant for the session tomorrow. When you change the process in a production of an antibody or when you go from a product to a biosimilar, the main changes that you may expect, not the distribution is allowed. distribution

only changes, are at the level of the glycosylation. There are some examples in the

pharmaceutical history where changes in production made a big impact in terms of pharmacokinetics and in vivo and dose potency, for example the famous XOMA’s 9180 antibody (note: RaptivaTM) for the treatment of psoriasis that was transferred

only. No further No only. to Genentech and had diﬀerent pharmacokinetics and potency. My question is

whether there are other examples that we should be aware of where changes in the glycosylation of antibodies result in diﬀerent pharmacokinetics, extravasation and in vivo potency.

vayinstitutes.be/ Ian Wilson: I am not sure I am the best person to answer that. I will transfer that PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- to Chi-Huey Wong.

Chi-Huey Wong: I think there are limited studies on that, so it would be interesting http://www.sol to see how that would effectively affect the behavior in vivo. But as I mentioned, if we take the known antibody, which is usually a mixture, and we analyze all the host on on host glycoforms and see how each of them behaves in animals, you will see that the sugars have a dramatic effect on ADCC (antibody-dependent cell-mediated cytotoxicity). If the antibody is against cancer and has also an anti-inflammatory activity, it is a chimera. I think the Fc receptor somehow interacts with the glycan. We need to http://www.worldscientific.com/worldscibooks/10.1142/ have a structure of the complex to better understand how the sugars of the antibody play a role in interacting with Fc γ receptor.

Richard Lerner: There is a sort of famous experiment. As you might imagine in the biosimilars arena, there were the “erythropoietin wars”. The Irish wanted to go into the business of biosimilars and so they built their production plants. What they found was that the commercial reference erythropoietin produced was more

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World diﬀerent batch to batch than their own erythropoietin. The good news for the

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY pharma companies was that under the new “biosim” rules, it is good to have a mess because you can’t replicate a mess. If you have a homogeneous thing, you can make it and it becomes like a small molecule. But if you put out a product, let’s say a peptide product, and there are many components for which you don’t quite know what the active principal is, even if they all have the same sequence, you cannot make a “biosim” unless you copy that complete mess. It is a very October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 160

160 Microbiomes and Carbohydrate Chemistry

interesting dynamic because in the small molecules world, one attempts to make the most pure compound you can get your hands on. In the “biosim” world, it would be interesting to hear from the people who worry about fermentation: can

you actually by fermentation get the same sort of carbohydrate derivitization twice

in row?

Chi-Huey Wong: I mentioned about the N-glycosylation from yeast to human. So the core trisaccharide would be the same from yeast to human. If we take a glycoprotein made from yeast to create the sugar to keep the (trimannose) core and then add the one of interest to the core, we create the human version of the distribution is allowed. distribution

glycoprotein. But it depends on if you want a mixture or a single molecule. . .

Richard Lerner: But this is because you developed an elegant method! What would

9180 you expect is going on with all the other products out there? only. No further No only. Chi-Huey Wong: Maybe Markus (Aebi) will answer that. I think my expectations are in the future. Biosimilars will be one business, but people will try to get a single molecule out of biosimilars as a new drug. That would be a new drug instead of a biosimilar. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Markus Aebi: I think for a single site like a protein this is possible. You can mimic this microheterogeneity, but as soon as you go to multiple sites, this becomes very difficult and the reason for that is that we lacked until now analytical tools to http://www.sol quantify structures per site. I think now we have these tools and this means that we can start addressing the question: what is the function of a specific carbohydrate host on on host on a specific site? Until now we always had handwaving: it has no effect, but we didn’t have the tools to generate glycoproteins with a defined structure at the defined sites. Now we can start asking these questions and I think this will be important in the future. http://www.worldscientific.com/worldscibooks/10.1142/ Peter Seeberger: Along those lines, Danishefsky et al. have just finished the first total synthesis of chemically made erythropoietin. This is entirely synthetic and basically made like a pharmaceutical drug. It is a single molecule, completely defined. I wouldn’t say it is a drug at this moment because it is too expensive to produce, but it gives us a chance to really understand at the atomic level what is the exact composition. Then you can find of a new product with a composition

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World based on that. I think the delineation between biologicals and small molecules is starting to fade. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Richard Lerner: We could have an endless discussion on this. It is true, scientiﬁ-

cally speaking, with modern analytical chemistry, you can really characterize your product down to the level of the proton. But it is a practical issue, and scientiﬁc issues and FDA practical issues are diﬀerent. And the FDA says that biosimilars must be the same, and that is just the way it is. When you start to ferment multiple October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 161

Session 3: Discussions 161

kilogram quantities of a protein like EPO, you cannot have two batches that are the same, it just doesn’t work that way. That, in the end, is the real issue. We, as scientists, tend to go for purity, while the regulatories (FDA) go for sameness.

Tom Muir: I have a question for the microbiome folks, maybe Dennis would be the

best person to answer that question. I remember my late colleague Ralph Steinman often told me that by many measures the gut is the largest immune organ in the body. It is stuﬀed full of dendritic cells, T-cells and so forth that keep a certain level of activity all the time, it is not quiescent. There is activity to some level and the idea is that there must be some kind of communication between the microbiome distribution is allowed. distribution

and the immune system to maintain homeostasis. But as far as I know, we don’t

know what the signaling molecules are. It seems to me that it might be a really interesting class of molecules to try to identify using methods such as the ones you 9180 talked about. Can you comment on that? only. No further No only. Dennis Wolan: You are absolutely right. The roles that the levels of the immune system has and the basal levels that are there is brought about by the commensal healthy microbiome. And exactly what kind of metabolites and/or interactions or peptide interactions are being sampled by dendritic cells in the Peyer’s patches vayinstitutes.be/ and things of that nature are still in progress. As these become elucidated, we PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- will deﬁnitely be using those as starting point to identify bacterial proteins and/or human expressed proteins that interact with them. http://www.sol Richard Lerner: Tom, you could put a really ﬁne point under that question. When you have something like a mycoplasma, that is an obligatory parasite, it has to live host on on host on the surface of our cells for life and escape the immune system. A) How does it do it? B) If we knew how it does it, could we use that strategy to make immunosup-

http://www.worldscientific.com/worldscibooks/10.1142/ pressive drugs? Because apparently as far as we know certain strains of mycoplasma will live inside you, for life, and totally exposed.

Brian Roth: Related to that question, it looks like this would be a tremendous opportunity for sort of unbiased small molecule based screening; if collections of these can be obtained, they can be screened in a high-throughput fashion against

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World families of proteins. We had a small collaboration with some group at UCSF, which

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY was sending us small collections. But it would be great to get larger collections of thousands (if not potentially millions) of these compounds secreted by bacteria that potentially aﬀect human gut homeostasis and immune cells and so on.

Dennis Wolan: There are absolutely groups that are focused on identifying the metabolites that are produced by both the bacteria as well as the human, and the interactions that result. Michael Fischbach from UCSF for example is trying to determine the natural products that are produced by the bacteria. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 162

162 Microbiomes and Carbohydrate Chemistry

Brian Roth: He had synthesized a small number. I can’t recall oﬀhand what the data are. This work might still be in progress. But yes, I am aware of that.

Markus Gr¨utter: I have a general question on practicality. How diﬀerent or similar is it to study the bacteria that exist for example in the gut and what is the diversity

and the (number of) diﬀerent strains. Does one have to study thousands of diﬀerent strains? Are there similarities? What is the situation there? I would be interested.

Bernard Henrissat: I can begin to answer your question. Everything depends on how we deﬁne a species to begin with. There is ample evidence of a large amount of horizontal transfer between bacteria in the dense environment that is the gut. And distribution is allowed. distribution

therefore you can imagine that if you have a bacterium that incorporates a piece of

DNA from another, its 16S RNA will not change. But the functional properties of

9180 that bacteria will change. So it really depends on how we deﬁne species. We try to understand and we would like to answer this question with the current knowledge, only. No further No only. which is not enough to derive everything. That is all I can say.

Markus Gr¨utter: But how about the practicality of working with these organisms/microbacteria compared to our laboratory animals? vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Adam Godzik: I think with there are a few strains of speciﬁc bacteria which emerge like a model system in in vitro studies that are being used. For the question related to the gene transfer, we don’t know how relevant these strains are to the counterparts http://www.sol in the real system. On one hand metagenomics will study the distribution of strains like we have in a normal gut. And on the other, we have model systems but there is host on on host still ongoing debate on how related they are. I know a few cases where for instance some models used did not incorporate interesting genes and therefore for in vitro studies turn out to be irrelevant. But we are just developing this library of relevant organisms that could be used for models. http://www.worldscientific.com/worldscibooks/10.1142/ Dennis Wolan: There is the clear example of a situation where you can’t just use a single model organism. B. theta (Bacteroides thetaiotaomicron) for instance, that Ian mentioned, encodes 400 or so glycosidases. And it is not until B. theta had been co-inhabited into a mouse with some probiotic (I don’t remember the name) that all the glycosidases become expressed. So the interactions between the individual bacteria and how they interact with their host is another layer of complexity. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Christine Orengo: Can I just ask another practical question? I think you alluded to NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY it before about those metadata. If we are trying to understand the functions that these communities have, it is very important to capture the phenotypes that we see

in either human health or human diseases or, if you are talking about soil bacteria, then pH and diﬀerent characteristics. Are there any initiatives to derive ontologies of the type of information that should be collected along with the sequencing data? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 163

Session 3: Discussions 163

Adam Godzik: The answer is yes, there is a consortium for developing standards for metadata from marine organisms. This was most related to microbiomes as I mentioned until this ﬁasco where all samples were collected without any information

about water temperature. It was very diﬃcult later to compare them to anything

because the samples arrived “naked”. For humans, it still doesn’t exist. We have problems analyzing certain data sets because we don’t know about the medical history of people or about antibiotic use for instance. I think at this point, the concept is to collect all information that is available, just put it in an unstructured way and worry about it later. distribution is allowed. distribution

Ian Wilson: I would like to move on and talk about synthesis of glycans. Is this now

conquered? Can we get gram quantities? Are there other sequences, in particular of ones that are put on in human cells? Can we get any amount pretty readily? 9180 Peter Seeberger: We can make up to kilogram quantities of certain glycans — only. No further No only. of certain mammalian type — but also of bacterial type. We typically on the instrument I showed you make 25 micromoles of material which is about somewhere between 10 and 50 milligrams of what you need. I think this is suﬃcient for most studies we would like to do. The question that of course comes up: Can you make all vayinstitutes.be/ structures? The answer to that is clearly no. Another question is how many building PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- blocks do we need to make for example a mammalian glycome? Theoretically, that is in the hundreds of structures but from bioinformatics studies, we found out that

http://www.sol we can get away with probably 15 building blocks to make most of the human glycome. When you go to bacterial glycomes, the number is larger. But within

host on on host certain types of bacteria you can get away again with very fewer glycan building blocks. So I think we are still ways away from making everything, but if you really want to make something, we can do that. I give the example of a compound called “Prevnar”, a 13-valent vaccine against Streptococcus pneumoniae. It sells itself for billions dollars worth from Pﬁzer every http://www.worldscientific.com/worldscibooks/10.1142/ year. We have now succeeded in making this entire thing using some synthetic chemistry. Of course the Streptococcus pneumoniae has more than 13 serotypes, there are 96 serotypes. We haven’t made all the 96 yet! It is a long way away, but if you really want a carbohydrate, you can make it. And I think like in DNA, people couldn’t make poly-G for many years. In peptides, you could not, after Merriﬁeld (pioneered solid-phase peptide synthesis) make everything right away. I think there is room for improvement, but again, if you really want something, I really do believe ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World you can make it. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Chi-Huey Wong: I agree. I think if there is a demand the synthesis can be solved. The question is: What to make? The information has to come from biologists.

Kurt W¨uthrich: As a follow up to Markus Aebi’s talk, are there any systematic studies of polypeptide chains with and without glycosylation? I mean unfolded October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 164

164 Microbiomes and Carbohydrate Chemistry

polypeptide chains and studies on their folding behavior? It may be diﬃcult to keep some of these polypeptides unfolded in solution, but it is feasible. Has anything of this been done?

Markus Aebi: I think there are folding studies on RNAse A and RNAse B but I am

not aware of others, I must say.

Chi-Huey Wong: I think there are many examples (from Patel, from Barbara Imperiali’s and many other groups). We had one from about 15 years ago: what we did was make the dodecamer of the RNA polymerase C-terminal repeat that is in random coil. But if you add the mono-sugar, it becomes a type-II β-turn. So distribution is allowed. distribution

the glycosylation is able to induce the conformation from random coil to type-II

β-turn. This of course will initiate the folding. But there are many examples using

9180 a peptide as a model, among which synthetic peptides with and without sugar to see how the glycosylation aﬀects the folding. only. No further No only.

Tom Muir: I think the erythropoietin Peter eluded to from the Danishefsky group, actually did see quite dramatic diﬀerences in the folding yield of the protein, as a function of how many sugars were added to it. And it wasn’t until they had the

vayinstitutes.be/ full complement on that the folding became reasonably eﬃcient. Without them, it PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- was really a mess due to the folding reaction. That is speciﬁc of the in vitro folding reaction.

http://www.sol Peter Seeberger: I think the answer is only as good as the models are. Early on those were relatively short peptides with relatively few sugars. And then it is diﬃcult to host on on host come to a general conclusion based on that. As we see larger molecules with larger glycans, we will get more information. Of course what this means biologically is less clear. Because some of these glycans are put on during the synthesis of polypeptides in some cases, there may be immediate eﬀects on that. So I think that answering

http://www.worldscientific.com/worldscibooks/10.1142/ to the question is not that straightforward.

Judith Klinman: I haven’t heard anything about lifetime of proteins in the cells as a function of glycosylation. It is something I remember from years ago but I don’t know what the current status is on that thinking.

Markus Aebi: So glycoproteins are secreted proteins, so the relevance is either on the plasma membrane or they are secreted in the serum. In the serum, we know very well ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World that the glycan structure determines the lifetime. The asialoglycoprotein receptor NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY that retrieves non-siaylated glycoproteins from the serum and thereby determines the half-life. On the plasma membrane it is an interesting question. How is the

quality controlled with respect to the glycocalyx on the cell surface? There are some ideas around of how this is managed and what determines the half-life. But until now we still don’t have the tools to analyze the half-life of a glycoprotein in vivo. This is not something you just do on a Saturday morning! October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 165

Session 3: Discussions 165

Markus Gr¨utter: As follow up on Kurt W¨uthrich’s question to Markus Aebi: It seems that you have the machinery studied and you know the components that are responsible for the glycosylation on the polypeptide chains. Do you see a possibility

to engineer these components sort of semi in vitro or directly glycosylate something?

Markus Aebi: So one way is you separate the two things, namely the N-glycosidic linkage, to make the N-glycosidic linkage from the structure of the carbohydrate. This is an engineering way. You can actually separate the two processes. Making the N-glycosidic linkage is at a speciﬁc point in the polypeptide chain. For this, you have to rely on enzymes; you cannot do that chemically except with big eﬀorts. So distribution is allowed. distribution

there are E. coli systems that can do that, you can then combine it with chemically

synthesized carbohydrate and ligate the two things together. That is possible, now whether it is commercially attractive I don’t know. I think as soon as the regulations 9180 are such that you require really pure components as we discussed before, then it

only. No further No only. becomes commercially interesting because then it can compete with the production

from eukaryotic cell lines.

Chi-Huey Wong: I have a question for Markus. You mentioned that it is maybe the kinetic barrier that gives the mixtures of glycans. You know glycosyl transferases vayinstitutes.be/ are membrane-bound in the Golgi. I wonder if there is lectin involved to move the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- newly synthesized glycoproteins around. Can you further elaborate on why we get mixtures? http://www.sol Markus Aebi: There is a Nobel prize just awarded for this transport phenomenon in the cell on this vesicular transport. There is still some debate as how proteins are host on on host transported in the Golgi, whether there is retrograde transport or recycling. I don’t want to go into this debate but it is clear that the protein cargo travels through this Golgi and the membrane-bound glycosyl transferases stay within the stacks creating individual gradients and therefore a microenvironment in these micro-reaction http://www.worldscientific.com/worldscibooks/10.1142/ vessels, so to say, that then determine the structure of the carbohydrates. That is at least my view. I think it’s a given set of glycosyl transferases in the cell that is the hardware. And then you determine the expression levels individually, and these determine in what direction glycosylation goes. And then the speed of transfer that then ends up in speciﬁc glycan structures on a protein. That is how I view it.

Ian Wilson: Can I ask you Markus about the site-speciﬁc identiﬁcation? Because

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World that is something I have been extremely interested in. This has really held our work

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY up very badly when trying to think of immunogen design for an HIV vaccine. It is complicated with HIV, due to the 27 glycans per monomer (that is 81 per trimer), so we really would like to know the speciﬁc glycans at particularly positions. How can we do that with such a large protein, because with each tryptic peptide you get an average of 4 glycans.

Markus Aebi: We are using analytical tools that were developed at the Academia Sinica to exactly address this question. These are MS-based analysis HCD fragmen- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 166

166 Microbiomes and Carbohydrate Chemistry

tation that allows you to identify ﬁrst glycopeptides and, at the same time from the same fragmentation, use the information to identify the polypeptides. Then we use as a second dimension sequencing of the carbohydrates. This new technology

allows us to exactly address this type of question and quantify the corresponding

results.

Ian Wilson: You were suggesting in your talk there wasn’t perhaps as much variation as one might think at a speciﬁc position and that there are actually relatively few isoforms/glycoforms there. Did you ever see mixed high mannose and complex sugars at the same position or is that never found? distribution is allowed. distribution

Markus Aebi: In this system (a purely experimental system), we are using yeast

proteins expressed in insect cells. This is as reductionism as it can be. There we did

9180 not see high mannose structures expressed except for those in the ER. But I would expect that as soon as you have very strong interactions between a carbohydrate only. No further No only. and the surface of a protein then you get high mannose structures on the ﬁnal protein. And I would be eager to measure these interactions. We can model it by molecular dynamics, but I think NMR would be an ideal system to see how speciﬁc these interactions actually are and whether we are able to predict them for the vayinstitutes.be/ outcome with pure protein. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Jason Chin: I have a question about glycosylation in the Golgi in two parts. You talked about a series of reactions happening as you move through the Golgi. And http://www.sol then possibly co-translational initial glycosylation, which possibly precedes the folding event. My ﬁrst question is about the co-translational folding: To what extent host on on host is it possible to recapitulate these things in vitro and to make detailed biophysical measurements? My second question is about the subsequent series of glycosylation and how we should think about those. Is there any directional ﬂow where you have a series of glycosylations and any error you make in any step is propagated to http://www.worldscientific.com/worldscibooks/10.1142/ the next? Or is it more sophisticated than that it that and there might be retrograde transport coupled to deglycosylation and error correction? How do we think about this process and how can the analytical techniques that we have allow us to distinguish between those possibilities?

Markus Aebi: I think ﬁrst that the term “co-translational” you used is relevant. It is more “before folding” or “after folding”. To couple the glycosylation to the

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World translocation event is one way to make sure that glycosylation occurs before folding.

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY But it can also occur much later on in the process, in the folding pathway. We cannot mimic, but we can do in vitro glycosylation and we couple it to the translocation event. These studies have been done extensively, also to monitor other things of the translocation events and glycosylation was used as a marker. We can detail how far the (poly)peptide has to reach into the ER lumen to get glycosylated and we know quite well from in vitro studies how this processes work. In terms of October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 167

Session 3: Discussions 167

retrograde transport, we take it oﬀ again. We lack to my knowledge the enzymatic machinery in the lumen of the ER to do that. To deglycosylate proteins, they do have to exit the ER again in retrograde translocation. Then they are degraded in

the proteasome. But before the sugar has to be taken oﬀ from the polypeptide.

Jason Chin: So this means there is no error correction?

Markus Aebi: No. You just degrade the protein that is not properly folded.

Jason Chin: OK. But you never correct the glycosylation on the protein? distribution is allowed. distribution Markus Aebi: As far as I know, no. Because taking of the sugar generates an

aspartic acid instead of an asparagine, which wouldn’t make much sense. 9180 Ian Wilson: Can I ask some questions about identifying the glycan speciﬁcity? The only. No further No only.

glycan arrays are fantastic inventions of the last few years in order to decide what the specificity of the glycan is. But we have encountered a number of situations where we actually don’t find much specificity on the array. The binding is just too weak. So you either need higher density arrays or you need mixed glycans. The particular examples that I showed you are antibodies that actually have 2 or 3 vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

binding sites for a glycan plus interaction with the polypeptide. So the individual - interactions are likely to be probably very weak (probably in the millimolar range), but can I ask about the detection and the speciﬁcity. How can you actually detect

http://www.sol so weakly binding glycans?

host on on host Peter Seeberger: I think all the data you get out of your glycan arrays is only as good as your input, as far as content of the array is concerned. I think we are only at the very beginning. The Consortium of Functional Glycomics I think has 600+ compounds. We have now hundreds of compounds on our arrays. But it depends on what you put on there. We cannot believe that with 600 glycans, we really http://www.worldscientific.com/worldscibooks/10.1142/ even begin to cover either the human glycome or the microbial glycome. And I think particularly the arrays that are out there right now are weak on the bacterial glycome. There are almost no bacterial structures on there. That is what my department has been addressing these recent years because they feel there is a real need. They are now doing the same thing for plants. So I think the content has to be improved, we need more glycans. It is not going to be pretty but someone has to make those molecules, or isolate them. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World The second question is the interactions between proteins and glycans. You are NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY implying most of these are weak interactions. I think this is only partially true. There are some known interactions where the binding is in the low nanomolar

level. Of course people have dogmatically written about it that this is not the case, but I think we see very tight binding, particularly in the case of glycosyl-glycans. Heparins and proteins bind exceptionally tightly to each other. There is charge, so these things are possible. I think it depends on what the glycan should do: if October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 168

168 Microbiomes and Carbohydrate Chemistry

the glycan is at the surface and it has to temporarily engage the receptor, it makes sense that we get a weak binding. But then binding is temporarily and it opens up again. If you want a signaling event in the cell, I think you want to have a

tight and speciﬁc binding event that leads to the transmission of information. And

I think that depending on what the function of the glycan is, we can have diﬀerent types of binding that range from very tight to not tight at all, where multivalency and other things plays a role, but I think people have been quick in making broad ranging statements about the functions of glycans based on one or two examples in some cases. So I do not say we do not need some sort of identity arrays, but I think we need both and we should be aware that in some cases with glycans we see very distribution is allowed. distribution tight binding.

Ian Wilson: I mean the situation I am talking about is: we actually know roughly 9180 what the speciﬁcity is, we know the sort of glycans that are present on the array,

only. No further No only. so that is not the issue. We know that we can get crystal structures. We actually

have sometimes crystal structures even with the individual glycans and with the glycosylated protein. We actually see these interactions that we are not detecting on the array, and we know that the glycans are actually on the array.

vayinstitutes.be/ Peter Seeberger: I think this is a matter of context. Because I think people have PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- focused only on the protein or only on the glycan. It would be really interesting to look at the whole system, because glycosylated proteins have both components:

http://www.sol the protein and the glycan. And I think it has been diﬃcult in the systems that have been made by Markus and others, and through synthetic means we are getting

host on on host towards this end now.

Chi-Huey Wong: I can add some points about this. The antibodies you discovered from HIV patients, some of them, like 2D12, PGT16, PGT9, they don’t recognize just one glycan. They recognize two diﬀerent glycans. How do you pick that http://www.worldscientific.com/worldscibooks/10.1142/ up from the array? The current array is made from glass slides activated by N- hydroxysuccinimide ester (NHS). We know it is unstable and very often you see that the active ester is not evenly distributed on the surface. So the array you have got is not very homogeneous. I showed that aluminium oxide-coated glass slides will be very even because the chemistry used for such arrays is diﬀerent. It is based on the use of a phosphonic acid tail, which will react with aluminium oxide spontaneously. That will create a much better controlled array with a pre-arranged ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World distance between the sugars. With that we are able to pick up the heteroligand

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY binding and multivalent binding with very high dissociation constants there. And I think this could lead to the design of carbohydrate-based vaccines in the future. The other problem with arrays, as Peter mentioned, is that we don’t know how

many sugars we have to put on. I think the NIH commission has several laboratories to make 10,000 glycans to be put on the array. But whether that is enough, we still don’t know. If we get into the microbiome system, it would be enormous. So October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 169

Session 3: Discussions 169

I think it is probably better to make the array for special purposes: we have the array for inﬂuenza, for cancer, for gp120 (HIV) (because it has about 85 diﬀerent glycans that are put on the array), and that is how we use it to understand the

speciﬁcities of the carbohydrate-binding antibodies.

Aled Edwards: I think this experiment must have been done? How much success has there been in reconstituting in S. cerevisiae by knocking them all out one by one, getting a minimal glycosyl transferase-ome and then adding back and looking what eﬀects they have. And are you going to do that with the baculovirus with the CRISPRs — start knocking out one by one so you can have an in vivo reconstituted distribution is allowed. distribution

system as it were?

Markus Aebi: I know from one example where they have engineered insect cells

9180 (I forgot the cell line) for the speciﬁc production of antibodies with fucosylated glycan structures, for which they have knocked out and put in the speciﬁc glycosyl only. No further No only. transferases. I think it was a Japanese company and they were able to generate a pure glycan structure on the produced antibody. So this is possible.

Aled Edwards: I was more thinking of reducing (I don’t know if there are 50 of

vayinstitutes.be/ them) down to ﬁve. The cells will be barely alive and then add them back (glycosyl PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- transferases) one by one to see exactly what happens.

Markus Aebi: So when you ﬁddle around with ER glycosylation machinery, you are http://www.sol dead! When you take the early enzymes in the Golgi, you hardly survive. And for the later ones, the sialyl transferases, there is no serious phenotype in mice. host on on host Chi-Huey Wong: The knock-out experiment was not very successful, particularly in mammalian cells, because they (glycosyl transferases) are all essential. One glycosyl transferase could be responsible for many diﬀerent glycoproteins. But one

http://www.worldscientific.com/worldscibooks/10.1142/ successful approach is the Genentech approach that uses a fucosyltransferase knock out to make antibodies without fucose. That has been produced on larger scales for development.

Ian Wilson: May I ask Chi-Huey Wong about the vaccine approach? A lot of the earliest thoughts on trying to get better interactions/responses for neutralizing anabolism, better immune response against viral antigens, was actually to cover up the surface with additional glycans, not to take glycans oﬀ. And then you actu- ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World ally only leave particular sites (such as receptor binding sites or fusion machinery) NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY unexposed, but you are covering up everything else including all the hypervariable loops. So you have done completely the opposite, that is you have taken all the

sugars oﬀ and you found you can get good responses. But assuming that you must get a lot of non-neutralizing response mixed in with neutralizing response, how do you boost the neutralizing response? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 170

170 Microbiomes and Carbohydrate Chemistry

Chi-Huey Wong: That is a good question. Where we have evidence so far is that once we remove the outer part of a sugar, you generate more diﬀerent kinds of antibodies. We used a single B-cell to get about 400 clones and some of them

are very interesting. We even obtained antibodies that recognize the sequence

covered by the sugars. I think we need to get a complex to better understand that. I think what happens is that (you know antibody-protein interactions are much stronger than antibody-glycan interactions) the antibody pushes the sugar away and thermodynamically you get the antibody-protein interaction there. We need to get that evidence by structure, but numerous studies showed that we get more diﬀerent kinds of antibodies and they target new epitopes. The antibodies distribution is allowed. distribution generated are better in terms of neutralization. We have to do more studies to

conﬁrm that.

9180 Richard Lerner: I think this is important — more than a gem of an idea — this

only. No further No only. dialogue between Ian Wilson and Chi-Huey Wong. Because the neutralizing an-

tibody people have pointed to the fusion area, there has been a race between the vaccine makers to sort of make the fusion piece, and jettison everything else. From a synthetic point of view, that turned out to be very diﬃcult to do, because once you take those pieces out, they don’t fold right. So the alternative would be to do vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS what Ian Wilson just said: that is leave the thing intact but cover everybody else

- with sugars. You have the cake and eat it too! You have got the epitope you want in the right conformation more or less, and you have blacked out everything else.

http://www.sol That would be a way to focus vaccination. You (to Chi-Huey Wong) could take the same technology you have had, turn it on its head and use it to cover everything. host on on host Chi-Huey Wong: I agree. We have just begun to understand how carbohydrates are taken up by B-cells and by dendritic cells, and how these antigens are processed, how they are cleaved and then presented to T-cells. This is unknown. We begin to understand because several carbohydrates make small changes there. What we http://www.worldscientific.com/worldscibooks/10.1142/ know is that in the case of proteins, of course protease is involved in the processing, but for carbohydrates we don’t know about it. People think it has to do with nitrogen oxide radicals cleaving glycosidic bonds but there are some other glycosylases involved. We don’t quite understand the speciﬁcity of that, and that is why often one carbohydrate by chance sometimes creates diﬀerent kinds of antibodies, which can recognize the internal part of the epitope.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Richard Lerner: But by the end of the day, the immune response is a binding 8 NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY energy competition amongst 10 combatants. And anything you can do to lower the binding energy potential of the rest of the molecule in terms of the initial event, never mind the protease later on, will focus the immune response. You can probably

focus on everything that you want.

Ian Wilson: There is evidence of what you are suggesting that certainly for these antibodies to the hemagglutinin, that the glycans are pushed out of the way. You October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 171

Session 3: Discussions 171

can see that the best antibodies are actually navigating through the glycans. If they weren’t there, you might get at the surface better, but if you think of trying to get an immunogen that is going to raise the same type of antibodies, you might

need to start with something that has less glycans on it, then add something later

as a boost with more glycans. You should use this combination approach. There is good evidence in the HIV field, this is also true, that you can start with something which has less glycans on it. There are usually less glycans on the transmitted virus than on the virus that ends up after a few rounds of replication in your body. Many times, the germline antibody doesn’t even seem to recognize the viral antigen. And one way that it sometimes can recognize is if you take the glycans off. There is distribution is allowed. distribution thought now of, again, taking some of the glycans off to get initial priming of the

germline response and then add more glycans back to make it more native-like.

9180 Richard Lerner: But nobody would have ever done that, if you had not shown that

only. No further No only. the loop can penetrate. That is what changes everything because why would you

take a sugar oﬀ to expose an epitope underneath, unless you knew what you know now. Because the natural system would cover it up again. You have to consider that, in the context of the idea, that there is enough energy in the loop as it were to push the sugars away and get at what is underneath. If you can’t do that, there’s vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS no sense doing it that way.

Kurt W¨uthrich: We all seem to agree that glycosylation is extremely important and

http://www.sol that we don’t know the basis for its important role. I have a speciﬁc example: Peter Seeberger talked about prion proteins and the GPI anchor. There is another story

host on on host related to the glycosylation of the prion protein in the cellular form. There are two glycosylation sites and based on the determination of the glycosylation levels, it has been determined that some people, mostly in England, were infected with a new variant of Creutzfeldt-Jacob disease. Just simply from the numbers, it seems that this may not have been a correct interpretation of the data. I can hardly http://www.worldscientific.com/worldscibooks/10.1142/ imagine that there would have been new variant of Creutzfeldt-Jacob disease in just 250 cases and never again, considering the extent of the BSE crisis in England. I suppose that since you worked on the GPI anchor site of the problem, you must have looked into this as well.

Peter Seeberger: Very simple answer: No, we did not yet. One of the limitations for us actually was to get the protein properly. You pointed out during the break ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World that it is not easy to work with this protein. I think the expertise of working both

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY with these complex proteins that also have to be handled taking into account certain safety regulations in some countries as well plus the required glycosylation expertise for us took a long time. So your plans now are that after we have made the protein

plus the GPI anchor to put on glycans as well. There is a group in my department that works only on that problem. They made progress but they haven’t yet made the whole thing. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 172

172 Microbiomes and Carbohydrate Chemistry

Ian Wilson: Do we have general questions that anybody wants to ask?

Christine Orengo: It is a change in topic completely. Just a brief question to Bernard

Henrissat: You talked about your subfamilies you are characterizing in the CAZyme classiﬁcation and you explicitly said you don’t use any sequence information. But

I imagine that within some families even from the tree you can see clustering and speciﬁcities associated with the diﬀerent branches. So I imagine there must be some sequence signals. I just wondered if you had a look at that at all and why you choose not to use that sequence information.

Bernard Henrissat: The situation is very diﬃcult. In a family-by-family case, we distribution is allowed. distribution

have ﬁgured out that there is no universal threshold to determine these subfamilies.

Sometimes in some families of glycosyl transferases, especially those that convert

9180 the blood groups A and B, just a couple of residues are enough to change speciﬁcity. In other families, we can go as low as 15% identity and still have the same function. only. No further No only. So really we need to have expert knowledge on each and every family. When we see subfamilies, we don’t know what is the driving force or whether we will discover new functions. They just provide opportunities for discoveries, but we don’t have a 100% hit on discovery. Most of the time we discover that what was the signal vayinstitutes.be/ turned out to be taxonomical drift, that sort of things. This is getting more and PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- more complicated because the trees get to be saturated by the sequence information we have these days. So it is easy to explain on a slide, it is a lot harder to do in

http://www.sol real life.

Don Hilvert: Maybe to return to a Kurt’s question about chemical synthesis of host on on host carbohydrates. What are the remaining chemical challenges to be able to make the entire universe of carbohydrates? Is it the building blocks or the particular linkages that are difficult to accomplish? Are there strategic changes that need to be made in the approach that one takes? http://www.worldscientific.com/worldscibooks/10.1142/ Peter Seeberger: Two things that you mentioned. The first is you need the building blocks. For human systems that is pretty easy, but for some bacterial systems that is difficult because you cannot pull out the monosaccharides. So we have to synthesize the monosaccharides from first principles. It is doable, but in some cases there are as many as 15 chemical steps required to get the building block. That has been worked out by a number of different laboratories. The more complex question is:

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World the chemical glycosylation reaction has been known for 110 years now. And to this

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY day, we know it depends on the electronics, sterics and conformation of both the nucleophile as well as the glycosylating agent. In many cases, we can tune that very nicely, but in some particular cases you get steric mismatches, and you do not get what you want. It has been so far diﬃcult to predict and so a few cognoscenti know how to do this and others don’t. We have tried to create an operator-independent database that compares many diﬀerent pairs of coupling partners and therefore derive some empirical evidence to predict glycosylations. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 173

Session 3: Discussions 173

This is growing now and I think this will make it easier for people to decide what building blocks to use. Based on that, we have come up with what we call a list of approved building blocks. We know they can be made, they are stable and

upon activation, and they give you reliably high activity at the site of glycosylation-

linkage. Their number is growing and I think this is what people who would like to make a glycan and are not experts need. They can either buy these building blocks and make the glycan themselves or go to a commercial supplier that will make a glycan for them. I strongly believe that the glyco-ﬁeld needs to integrate more with the people who make proteins because you cannot get a molecule pure, you can’t ask the (right) questions. Not everything can be made yet, but it can be made in distribution is allowed. distribution many areas.

Don Hilvert: On the other hand, the catalysts do exist in Nature. What about the 9180 chemo-enzymatic approaches? Is it just really the availability of the enzymes? Or

only. No further No only. the stability?

Chi-Huey Wong: I think that would be the way to go in the future. Of course it depends on the product. If you want to make glycoprotein or heparin sulfate or more complex structures, I would not just go around with chemical methods. If you vayinstitutes.be/ want to scale up, I think the enzymatic process could be the choice, because you PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- cannot tolerate any stereoisomer in the product, particularly if you are dealing with a vaccine. So I think both chemical and biological methods have to be considered

http://www.sol and of course it depends on what you want to make. The problem with the enzymes is that they are not available, but they are there! If there is a demand, I think they

host on on host will become available. This is very similar to the story of aspartame, the sweetener. When it came out, it was made by mixed anhydride chemical synthesis. The reason for that is that thermolysin was expensive and now it is as cheap as sodium chloride. When there is a big demand, I don’t think enzyme is a problem.

http://www.worldscientific.com/worldscibooks/10.1142/ Ian Wilson: A lot of what we’re having to do now is trying to think of not doing all glycoproteins but designing and synthesizing some sort of epitope scaﬀold. So you have maybe one peptide or multiple peptides and you want to put say Man5 in one place and Man8 in another place, and a complex sugar in another place. How can we do that?

Chi-Huey Wong: I think that this can be easily done by chemical means. Peter

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Seeberger is one who set up a solid-phase synthesis of target molecules. These are

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY small enough to be dealt with chemistry.

Peter Seeberger: I think one has to be pragmatic. As Chi-Huey Wong said before:

depending on the issue or what you have to make, you can use chemical means where you can create quickly diversity. If you go to large scales then you might do it enzymatically. But I think it is some of a “chicken egg” problem because materials are not available. People don’t even dare to ask about them. The technologies exist October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 174

174 Microbiomes and Carbohydrate Chemistry

but they are still expensive and people don’t dare to do it. At some point, you have to break from this vicious cycle.

Craig Townsend: I direct this question to the bio-informaticists. Is there any example of a drug that has its eﬀect against the primary target, but whose beneﬁcial

effects are amplified by normalizing in some way the microbiome? I understand it is hard to separate cause and effect here.

Adam Godzik: I am sure there are but I don’t know myself.

Dennis Wolan: There is the example that appeared in Science a few years ago: CPT- distribution is allowed. distribution 11 which is a chemotherapeutic that is processed by glucuronidases of the bacteria. So not necessarily what you are saying, but we have to take into consideration the

9180 metabolic processes on drugs in our PK-ADME studies for sure. only. No further No only. Ian Wilson: Any other burning question you want to ask to the panel? I am going to ask Markus one question: in HIV, we have this high mannose patch on gp120 and we are not quite sure how it arises but it seems to be conserved. Somehow the virus wants to hold on to this high mannose patch and we would like to understand

vayinstitutes.be/ why? Maybe for interactions with other molecules on the cell surface? How can PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- you control holding onto this high mannose patch with a hypervariable virus?

Markus Aebi: I believe it is relatively easy to maintain a high mannose patch be-

http://www.sol cause you simply have to prevent processing by mannosidases that are in the ER and in the Golgi. You have to hide the non-reducing end of the carbohydrate some-

host on on host how and you can do that by burying into a pocket of a protein. If that hypothesis is correct, we should be able to measure that by these interactions with a protein. I think the only way is to do that by NMR and more people have to work on this. I think it should be possible but Kurt might know more about if we could measure

http://www.worldscientific.com/worldscibooks/10.1142/ carbohydrate interactions with such a large molecule using NMR.

Kurt W¨uthrich: How large?

Markus Aebi: gp120?

Kurt W¨uthrich: No problem.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Markus Aebi: So you can model or you can measure. I would say it is the way to go. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Kurt W¨uthrich: Maybe this is a very na¨ıve question. Why did we exclusively talk

about glycosylated peptides or polypeptides and not at all about carbohydrates today? For example, carbohydrates as substrates for enzymes or the microbiome, especially in the gut? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 175

Session 3: Discussions 175

Bernard Henrissat: I did touch on this. I explained that we are only able to digest three sugars and the rest is entirely done by the microbiome. Lactose, starch and sucrose, that is the basic food and that is what gets absorbed by the small intestine

before it gets into the bowel.

Peter Seeberger: I think also it is interesting to see how these bioﬁlms form. What glycans are involved? How do they interact and how do the communities that live together in the diﬀerent parts of the body, how do they decide who is interacting with whom. I think glycans and carbohydrates are involved in these interactions. The problem, at least from my perspective, is that nobody knows the answer to distribution is allowed. distribution

that and I am not sure we actually have the tools at this moment to resolve this

sort of question. But to me that’s the real key question: why do certain types of bacteria decide to be in a certain place and deal with each other. And how do 9180 they communicate? Are glycans involved in that? Probably yes partially. Maybe

only. No further No only. someone has the answer here? I don’t think anybody has.

Markus Aebi: Kurt, you are absolutely right. You are touching a problem that glycans are not only here to decorate proteins. They are part of the cell wall, of the fungal and plant cells wall. You name it. But this is the general rule that vayinstitutes.be/ carbohydrates are made to interact with the outside of a cell and so you can touch PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- whatever you like. You will end up with carbohydrates at least in our view here. This is a totally new ﬁeld. We are the wrong experts. http://www.sol Chi-Huey Wong: I have a question related to high mannose cluster in human viruses. I don’t know why most human viruses have glycoproteins with the high mannose host on on host type: hepatitis, HIV, dengi? Does this high mannose glycan escape from the ER and never get into Golgi? Or do they go the Golgi and for some reason are not further glycosylated?

http://www.worldscientific.com/worldscibooks/10.1142/ Markus Aebi: I am not an immunologist but in my view, high mannose is “self”, a self entity recognized by the immune system. (N. Callewaert, you might correct me)

Nico Callewaert: The answer is multiple fold. HCV for example is packaged from the ER membrane directly. It doesn’t do that from the plasma membrane. The HIV does, as far as we know, but there are many viral glycoproteins that are

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World densely packed with one another, so steric occlusion away from the Golgi glycosyl transferases after that packaging has happened, maybe an important factor in that. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Ian Wilson: I think we have come to the end of this session. I thank the speakers

and panel members, and all of you for your questions. July 25, 2013 17:28 WSPC - Proceedings Trim Size: 9.75in x 6.5in icmp12-master

This page intentionally left blank distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 177

Session 4 GPCRs and Transporters: Ligands, Cofactors, distribution is allowed. distribution Drug Development

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 178

distribution is allowed. distribution

Phylogenetic tree of GPCRs and published GPCR structures. See

9180 Figure 1 contributed by Raymond Stevens on page 187. only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 179

179

GPCRs AND TRANSPORTERS: LIGANDS, COFACTORS, DRUG DEVELOPMENT

GUNNAR VON HEIJNE Department of Biochemistry and Biophysics, Stockholm University 106 91 Stockholm, Sweden distribution is allowed. distribution

My view of the present state of research in the ﬁeld 9180 We are currently witnessing an almost exponential increase in structural studies of only. No further No only.

integral membrane proteins. While a low-resolution structure of bacteriorhodopsin, obtained by electron crystallography, was published as early as 1975 [1], the ﬁrst high-resolution X-ray structure of an integral membrane protein, the photosynthetic reaction center, was published only in 1985 [2]. For many years there was but a

vayinstitutes.be/ trickle of new structures coming in — as late as 2000, there was only about 50 unique PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

structures in the PDB. Today, however, more than 400 unique structures have been - published (see http://blanco.biomol.uci.edu/mpstruc/) and new ones appear at an increasing rate.

http://www.sol What lies behind this small explosion of membrane protein structures? Many more crystallographers are active in the ﬁeld today than 10 years ago, and a number

host on on host of methodological advances have increased the success rate of structure determination. Perhaps the most important ﬁrst breakthrough was the adoption of a structural genomics approach: instead of working for decades on a single protein, most researchers today start from a large collection of homologous proteins collected from

http://www.worldscientific.com/worldscibooks/10.1142/ diﬀerent organisms (mainly prokaryotic, but now also eukaryotic). The genes are cloned, proteins are expressed in easy-to-use organisms such as E. coli or yeast, and rapid screens to single out those proteins that are produced in high amounts and behave well during puriﬁcation are applied. Finally, the best-behaved candidates are subjected to high-throughput crystallization screens in a wide range of detergents or lipid cubic phases, possibly with the addition of various ligands, lipids, or antibodies to help stabilize the protein. Proteins can also be stabilized by mutagen-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World esis strategies. Finally, crystals are screened at synchrotrons, and phases are solved either by molecular replacement (if a similar structure is already known), by the NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY introduction selenomethionine residues into the protein, or by classical heavy-atom derivatives.

But not all interesting membrane proteins have bacterial homologs. In such cases, the only open avenue is to tackle the diﬃculties of working with eukaryotic membrane proteins head-on. This realization has given rise to a dramatic improvement in techniques, both in terms of high-throughput approaches and in diﬀerent October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 180

180 G. von Heijne

ways of stabilizing these often ﬁckle proteins to increase the probability that they will crystallize. In parallel, new methodology has increased the number of membrane protein

structures solved by NMR — a technique that also allows studies of protein dy-

namics — and there are also new structures being determined by electron crystallography of two-dimensional crystals or by single-particle cryo electron microscopy. This session is devoted to two particularly important classes of integral membrane proteins: G protein-coupled receptors (GPCRs) and small-molecule transporters. Both classes contain centrally important drug targets, and play critical roles in many signaling and metabolic pathways. Both ﬁelds have seen great progress distribution is allowed. distribution during the past few years, opening up new possibilities for drug screening and

design. 9180 only. No further No only. My recent research contributions to the ﬁeld My own research deals with the question of how integral membrane proteins are targeted to, integrated into, and fold in the membrane during biosynthesis. These processes are common to all membrane proteins (including GPCRs and transporters), vayinstitutes.be/ and look very much the same in prokaryotic and eukaryotic cells. In short, ribo- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- somes translating a membrane protein are quickly targeted to so-called translocons, protein-conducting channels found in, e.g., the cytoplasmic membrane of bacteria, the endoplasmic reticular membrane in eukaryotic cells, and the inner membrane of http://www.sol mitochondria. The translocon then mediates the orderly membrane integration of hydrophobic transmembrane segments as they emerge from the ribosome. host on on host We have focused primarily on the energetics of the membrane-insertion process as it is played out in live cells. Recent contributions include detailed measurements of how much each of the twenty natural amino acids (and some non-natural amino acids to boot) contributes to the overall free energy of membrane insertion of a http://www.worldscientific.com/worldscibooks/10.1142/ transmembrane segment [3–5], as well as measurements of the forces exerted on the nascent polypeptide chain during membrane integration [6, 7].

Outlook to future developments of research in the ﬁeld Given the rapid pace of structure determination, representative structures for all

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World major subclasses of GPCRs and small-molecule transporters will no doubt become available within a few years. The next step is to obtain structural information on NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY protein complexes between GPCRs and other proteins; indeed, a structure of a GPCR in complex with its cognate G protein has already been solved [8]. Beyond

the static structures, the conformational changes involved in GPCR signaling and small-molecule transport will become increasingly well understood, thanks to structural, biochemical, and biophysical studies. New drug candidates will be developed, based on high-throughput screening approaches and structure-guided design. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 181

GPCRs and Transporters: Ligands, Cofactors, Drug Development 181

Acknowledgments Grants from the European Research Council, the Swedish Foundation for Strate- gic Research, the Swedish Research Council, and the Knut and Alice Wallenberg

Foundation are gratefully acknowledged.

References 1.R.Henderson,P.N.T.Unwin,Nature 257, 28 (1975). 2. J. Deisenhofer, O. Epp, K. Miki, R. Huber, H. Michel, Nature 318, 618 (1985). 3. T. Hessa, H. Kim, K. Bihlmaier, C. Lundin, J. Boekel et al., Nature 433, 377 distribution is allowed. distribution (2005).

4.T.Hessa,N.M.Meindl-Beinker,A.Bernsel,H.Kim,Y.Satoet al., Nature 450,

9180 1026 (2007). 5. K. Ojemalm,¨ T. Higuchi, Y. Jiang, U.¨ Langel, I. Nilsson et al., Proc. Natl. Acad. only. No further No only. Sci. USA 108, E359 (2011). 6. F. Cymer, G. von Heijne, Proc. Natl. Acad. Sci. USA 110, 14640 (2013). 7. N. Ismail, R. Hedman, N. Schiller, G. von Heijne, Nat. Struct. Mol. Biol. 19, 1018 (2012).

vayinstitutes.be/ 8. S. G. Rasmussen, B. T. DeVree, Y. Zou, A. C. Kruse, K. Y. Chung et al., Nature PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 477, 549 (2011). http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 182

182

STUDIES OF GPCR CONFORMATIONS IN NON-CRYSTALLINE MILIEUS

KURT WUTHRICH¨ Cecil H. and Ida M. Green Professor of Structural Biology The Scripps Research Institute, La Jolla, CA 92037, USA, and Professor of Biophysics, ETH Z¨urich, Z¨urich, Switzerland distribution is allowed. distribution

9180 My view of the present state of structural biology with GPCRs only. No further No only. The main goal of GPCR structural biology is to obtain insight into the mechanisms by which signals elicited by ligands in the extracellular “orthosteric” binding site are transmitted over a distance of about 35 A˚ to intracellular partner proteins (Fig. 1). This includes that one tries to establish correlations between the chemical structures

vayinstitutes.be/ of drug molecules and the signaling pathways, as well as to gain insight into the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- mechanisms of action of allosteric effectors, as indicated in Fig. 1. Present-day GPCR structural biology can work from a highly promising, recently established platform, since crystal structures of more than 20 different human GPCRs are http://www.sol available. These include the first structure of a human GPCR, β2AR, which was determined in 2007 [1], a high-resolution structure of A2AR[2],aβ2AR–G-protein host on on host complex [3], and numerous GPCR complexes with pharmaceutical ligands (surveyed in [4]). The availability of the high-resolution GPCR scaffolds provided by X-ray crystallography enables precise design of complementary experiments for studies in non-crystalline milieus, for example, with NMR spectroscopy or single-molecule

http://www.worldscientific.com/worldscibooks/10.1142/ spectroscopy.

Our recent contributions to GPCR research Using NMR spectroscopy in solution we pursued two diﬀerent lines of research. Firstly, in support of subsequent structural studies, we characterized solutions of detergent-reconstituted integral membrane proteins (IMP). This included the

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World screening of a wide array of newly designed detergents [5], the use of microcoil NMR equipment for measurements of translational diﬀusion coeﬃcients to charac- NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY terize the size of the IMP-containing particles [6], and applications of the results from [5] and [6] to optimize IMP solutions for NMR studies and for crystallization

trials [7–9]. Secondly, we inserted ﬂuorine-containing labels in strategic locations of GPCRs (Fig. 1). Observation of the NMR signals of these 19F-NMR probes enabled the observation of conformational equilibria between long-lived inactive and activated states of GPCRs, as illustrated in Fig. 2 for β2AR. This resulted in the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 183

Studies of GPCR Conformations in Non-Crystalline Milieus 183

distribution is allowed. distribution

9180 only. No further No only.

Fig. 1. Survey of GPCR structure and function. GPCR structures are characterized by seven trans-membrane helices. On the periplasmic surface there is an “orthosteric” ligand binding site (the arrow indicates mobility in the ligand binding site). Allosteric binding sites have been located along the periphery of the seven-helix bundle (cholesterol) and in the center of the bundle (Na+).

vayinstitutes.be/ Binding of drug molecules to the orthosteric site elicits signaling to partner proteins in the cyto- β

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS plasm, for example, G protein and -arrestin. The two red circles indicate locations for ﬂuorine-19

- labels that have been used for 19F-NMR observation (Fig. 2) of conformational changes associated with trans-membrane signaling between the orthosteric site ligand binding and the cytoplasmic protein surface. (Reproduced from [10]). http://www.sol

demonstration that signal transfer to G proteins and to β-arrestin are transmitted host on on host through diﬀerent signaling pathways [11, 12].

Outlook to future developments of NMR spectroscopy with GPCRs http://www.worldscientific.com/worldscibooks/10.1142/ Exploratory experiments with several diﬀerent GPCRs showed that the results on conformational transitions between long-lived active and inactive states, as obtained with β2AR, cannot readily be generalized for other receptors. It will therefore be exciting to pursue comparative studies of diﬀerent GPCRs with the 19F-NMR probe approach. In addition, we have established an expression system that enables labeling of GPCRs with stable isotopes, so that the 19F-NMR probe experiments can

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World be followed up with high resolution structural information, which may include de novo structure determination of new GPCRs. For me it is particularly exciting NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY that the glucagon receptor in its interactions with glucagon is one of our primary targets. Several decades ago, NMR studies of glucagon in aqueous solution [13] and

in the water/lipid interface on the surface of detergent micelles [14, 15] yielded interesting observations on the conformations of this polypeptide hormone in diﬀerent milieus. It will be exciting to compare these data with new results on the behavior of glucagon interacting with its receptor. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 184

184 K. W¨uthrich

distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 19 19 TET Fig. 2. F-NMR observation of conformational equilibria in β2AR. F-NMR signals of C265 TET and C327 in the apo-form and in four drug complexes of single-residue TET-labeled β2AR recorded at 280K. The experimental spectra (thin black line showing noise) have been deconvoluted http://www.sol into signals of an activated state (A, red) and an inactive state (I, blue) of β2AR. The thick black line represents the sum of the signals I and A. The chemical structures of the bound ligands are

host on on host schematically drawn on the right. (Reproduced from [10]).

Acknowledgments GPCR research in my laboratory is supported by the NIH Common Fund grant P50

http://www.worldscientific.com/worldscibooks/10.1142/ GM073197 and the NIH PSI:Biology grant U54 GM094618. I thank the colleagues and research associates listed in the references 4 to 15 for their courageous and devoted eﬀorts in our membrane protein research program.

References 1. V. Cherezov, D. M. Rosenbaum, M. A. Hanson, S. G. Rasmussen, F. S. Thian

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World et al., Science 318, 1258 (2007). 2. W. Liu, E. Chun, A. A. Thompson, P. Chubukov, F. Xu et al., Science 337, NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 232 (2012). 3. S. G. Rasmussen, B. T. DeVree, Y. Zou, A. C. Kurse, K. Y. Chung et al., Nature

477, 549 (2011). 4. R. Stevens, V. Cherezov, V. Katritch, R. Abagyan, P. Kuhn et al., Nat. Rev. Drug Discov. 12, 1 (2013). 5. Q. Zhang, R. Horst, M. Geralt, X. Ma, W. Hong et al., J. Amer. Chem. Soc. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 185

Studies of GPCR Conformations in Non-Crystalline Milieus 185

130, 7357 (2008). 6. P. Stanczak, R. Horst, P. Serrano, K. W¨uthrich, J. Amer. Chem. Soc. 131, 18450 (2009).

7. R. Horst, A. L. Horwich, K. W¨uthrich, J. Amer. Chem. Soc. 20, 16354 (2011).

8. R. Horst, P. Stanczak, P. Serrano, K. Wüthrich, J. Phys. Chem. B 116, 6775 (2012). 9. R. Horst, P. Stanczak, R. C. Stevens, K. Wüthrich, Angew. Chem. Int. Ed. 52, 331 (2013). 10. T. Didenko, J. J. Liu, R. Horst, R. C. Stevens, K. Wüthrich, Curr. Opin. Struct. Biol. 23, 1 (2013). distribution is allowed. distribution 11. J. J. Liu, R. Horst, V. Katritch, R. C. Stevens, K. Wüthrich, Science 335, 1106

(2012). 12. R. Horst, J. J. Liu, R. C. Stevens, K. Wüthrich, Angew. Chem. Int. Ed. 52, 9180 331 (2013). only. No further No only. 13. C. Bösch, A. Bundi, M. Oppliger, K. Wüthrich, Eur. J. Biochem. 91, 209 (1978). 14. W. Braun, C. Bösch,L.R.Brown,N.G¯o, K. Wüthrich, Biochim. Biophys. Acta 667, 377 (1981).

vayinstitutes.be/ 15. W. Braun, G. Wider, K. H. Lee, K. W¨uthrich, J. Mol. Biol. 169, 921 (1983). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 186

186

THE SEVEN TRANSMEMBRANE SUPERFAMILY

RAYMOND C. STEVENS Department of Integrative Structural and Computational Biology, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037, USA and the iHuman Institute, ShanghaiTech University, Shanghai, China distribution is allowed. distribution

My view of the present state of research on the 7TM superfamily

9180 (GPCRs) The 7-transmembrane (7TM) superfamily is the largest membrane protein family only. No further No only. in the human genome, and the target of more than 30% of all therapeutic drugs that treat diseases aﬄicting mankind. At more than 800 members and responsible for 80% of all human cell signaling, the family includes well-known receptors often referred to as G protein-coupled receptors (GPCRs). In 1976, Henderson and Un- vayinstitutes.be/ win were the ﬁrst to observe the 7TM protein architecture with bacteriorhodopsin PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- [1]. Palcewski and co-workers then reported structure-function studies of bovine rhodopsin in 2000 [2]. In 2007, structures of human β2-adrenergic receptor (β2AR)

http://www.sol were determined by Kobilka, Schertler, and Stevens combining different approaches highlighting the first structures of a human GPCR and the first receptor controlled

host on on host by diffusible ligands [3–5]. Since then, structures of more than 24 different 7TM receptors have been determined [6], each one impacting a different important area for basic and applied discovery. Furthermore, the structure of a GPCR G protein complex has been determined with an intracellular antibody [7]. At the current pace, it is highly likely that representative structures of all subfamilies will be de- http://www.worldscientific.com/worldscibooks/10.1142/ termined in the coming few years enabling new discoveries in molecular recognition and illuminating the incredible diversity of these receptors. They should also provide deeper insights into structural basis for activation. Collectively, the structures and related follow-up biochemical and biological studies are dramatically changing how we think about the 7TM receptors. For example, with the observations of cholesterol and sodium binding to receptors and complemented by functional studies, we are now starting to view these molecules as allosteric machines controlled ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World by many different molecules in the membrane beside just endogenous or synthetic NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY pharmacological molecules [8].

My recent contributions to research on the 7TM superfamily

(GPCRs) Like many others in the ﬁeld, my laboratory started pursuing structure-function studies of this superfamily more than 20 years ago, with no success until fairly re- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 187

The Seven Transmembrane Superfamily 187

cently. Combining our novel membrane protein technology developments created speciﬁcally for GPCR structure-function studies including miniaturization and au- tomation, alongside Brian Kobilka’s extensive knowledge of the protein chemistry of

β2AR, the structure of β2AR was determined at a resolution that enabled the char-

acterization of correct ligand orientation and many of the receptors novel features. Given our large investments in technology development, we were able to quickly follow-up with the structure of the human A2A adenosine receptor (A2AAR) [9], and also establish the human GPCR Network (http://gpcr.scripps.edu) that has successfully breached barriers to routine structural studies of human GPCRs [10]. Eﬃcient determination of GPCR structures requires the development of a robust distribution is allowed. distribution approach to enable a high likelihood of success and to provide a platform that

is amenable to optimization and cost reduction based on accumulated experience. The approach used by the Scripps GPCR Network in the initial successful structure 9180 determinations of the β2AR and the A2AAR led to the development of a pipeline only. No further No only. infrastructure with feedback loops that could be used as a template for the determination of other GPCR structures. This pipeline is constantly being optimized using a ‘family learning approach’, which is similar to the approach used by the Structural Genomics Consortium [10] most notably with protein kinases [11]. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

Fig. 1. Phylogenetic tree of GPCRs and published GPCR structures (24), those human GPCR structures published by the Scripps GPCR Network (16) are shown with red circles. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 188

188 R. C. Stevens

Since 2007, our process has determined the structures of two thirds of the available unique receptor structures (16 of the 25) and provided follow up insight into their structures and functions (Fig. 1). Each of the structures has been studied in

collaboration with a scientiﬁc community leader who has extensive knowledge of

the specific receptor system from a chemical, biological and pharmacological perspective. This highly collaborative approach is beneficial not only to ensure that the science is at the highest quality, and that insightful follow-up experiment and discovery are done, but it also establishes an avenue for extensive dissemination and communication of the data and results beyond the structural biology community. In addition, computational docking challenges with the community, specifically GPCR distribution is allowed. distribution Dock 2008 [12], 2010 [13], and 2013 are also helping to monitor the field of compu-

tational GPCR modeling and docking, to provide us with feedback for prioritizing target selection and thus ensuring continued high impact to the biomedical com- 9180 munity. only. No further No only.

Diversity and modularity of GPCR structures [6, 14] Although all GPCRs are characterized by a similar 7TM topology, the ﬁve major families of human GPCRs share little sequence identity (SI) and possess diﬀerent vayinstitutes.be/ <

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS extracellular N-terminal domains (SI 10% in the 7TM domain). The largest and

- most diverse Rhodopsin family (also called Class A) consists of approximately 700 GPCRs in humans, and its subgroups α–δ have an SI of 25% or more. Each sub-

http://www.sol group contains numerous subfamilies, and GPCR subtypes within these subfamilies share a higher SI (≥ 30%) and often a common ligand selectivity. A survey of all

host on on host solved GPCR structures is improving our understanding (and correcting misunder- standings) of the architecture of these receptors. GPCRs exhibit modularity in the extracellular module, which is responsible for binding diverse ligands and has much http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Fig. 2. General architecture and modularity of GPCRs. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 189

The Seven Transmembrane Superfamily 189

higher structural diversity than the intracellular module, which is involved in binding a limited number of downstream eﬀectors including G proteins, arrestins, and kinases and hence shows higher sequence/structure conservation (Fig. 2).

A highly diverse repertoire of structural features in the ligand-binding pockets of

diﬀerent GPCR subfamilies apparently reﬂects evolutionary pressure to selectively recognize ligands that vary greatly in shapes, sizes and electrostatic properties. Even within the preserved overall 7TM bundle architecture, there is a remarkable diversity of shapes and features of binding pockets between GPCR subfamilies, manifested in both side-chain diversity and variations in backbone conformations of TM helices and extracellular loops. distribution is allowed. distribution As an example, the structures of opioid and serotonin receptors involved in

binding of psychedelic drugs have been studied. In collaboration with Bryan Roth at UNC, these structures highlight the diversity in the binding pockets and now provide 9180 a molecular basis for answering the question as to why the nociceptin receptor only. No further No only. (NOP) does not bind morphinans, unlike the other classical opioid receptors, or how molecules like LSD elicit distinct and biased responses by closely related serotonin receptors. The structures show that replacement of only a few key residues in the ligand binding pocket region results in conformational changes in the structure

vayinstitutes.be/ including shifts in helices V/VI leading to diﬀerent pharmacological behavior. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Crystal structures of GPCRs also provide a robust three-dimensional structural framework for computational modeling of receptor dynamics and oligomerization state, as well as for ligand docking and virtual ligand screening (VLS). The growing http://www.sol number of structure-based VLS studies demonstrates encouragingly high hit rates (20–70%) in the identiﬁcation of new ligand chemotypes as lead compounds for host on on host A2AAR, CXCR4 chemokine, D3 dopamine, and H1 histamine receptors, as well as in lead optimization. Following their successful applications to kinases, proteases and other protein target families, structure-based ligand screening technologies are now becoming an important part of the GPCR drug discovery process in phar-

http://www.worldscientific.com/worldscibooks/10.1142/ maceutical and biotechnology companies. For example, the S1P1 structure [15] is now being used by the biotech startup Receptos in their drug development eﬀorts targeting multiple sclerosis and irritable bowel syndrome with two diﬀerent Phase II/III clinical trials currently underway.

7TMs as allosteric machines [8]

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Pharmacological responses of GPCRs can be ﬁne-tuned by allosteric modulators. Structural studies providing such observations have been limited by the medium NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY resolution of past GPCR structures. Thus, we reengineered the human A2AAR by replacing its third intracellular loop with apocytochrome b562RIL and solved its

structure to 1.8 A˚ resolution. This high-resolution structure allowed us to identify 57 ordered water molecules inside the receptor comprising three major clusters. The central cluster harbors a putative sodium ion bound to the highly conserved aspartate residue Asp2.50. Additionally, two cholesterol moieties stabilize the con- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 190

190 R. C. Stevens

formation of helix VI. These high-resolution details shed light on the role of water molecules, sodium ions, and lipids/cholesterol in GPCR stabilization and function.

Biased signaling pathways in 7TMs characterized by 19 F-NMR [16] Extracellular ligand binding to GPCRs modulates G protein and β-arrestin signaling by changing the conformational states of the cytoplasmic region of the receptor. In collaboration with the W¨uthrich laboratory at TSRI, we placed site-speciﬁc 19 F-NMR labels in the β2AR in complexes with various ligands, and observed that

distribution is allowed. distribution the cytoplasmic ends of helices VI and VII adopt two major conformational states.

Changes in the NMR signals reveal that agonist binding primarily shifts the equilib-

rium toward the G protein–specific active state of helix VI. In contrast, β-arrestin– 9180 biased ligands predominantly impact the conformational states of helix VII. The selective effects of different ligands on the conformational equilibria involving he- only. No further No only.

lices VI and VII provide insights into the long-range structural plasticity of β2AR in partial and biased agonist signaling. The use of 19F-NMR may thus prove to be a widely applicable tool for structure-activity relationship studies of GPCR ligands. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

Outlook for research on 7TM superfamily (GPCRs) - Over the past 40 years, pharmacologists have treated GPCRs as black boxes that recognize and respond to certain ligands, and narrowly deﬁned them as either cou- http://www.sol pled to G proteins, or more recently coupled to arrestin. However, we now know there are more than 200 other intracellular proteins that interact with the 7TM re- host on on host ceptor family and we are only beginning to understand signaling. Furthermore, with only 24 known structures, this represents only a few percent of the total number of receptors, each one characterized revealing great new discoveries no one expected. Our ultimate goal is to understand how complexes, cells, organs, and species

http://www.worldscientific.com/worldscibooks/10.1142/ work at a molecular level. To accomplish this endeavor and better model how organisms work, wide ranging types of data need to be integrated together including computational approaches. Human cell signaling with 7TMs at the center, are a model system to investigate, with broad reaching impact on many diﬀerent ﬁelds of basic and applied science to help mankind.

Acknowledgments ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World This work was funded by the NIH Common Fund for Structural Biology grant P50 NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY GM073197 and NIH PSI:Biology grant U54 GM094618. This work could not have been accomplished without the incredible teamwork eﬀort by Enrique Abola, Vadim

Cherezov, Vsevolod Katritch, Peter Kuhn, Angela Walker, Kurt W¨uthrich and all of the students and postdocs we have worked with over the past 20 years. I am also very grateful to our collaborators Ruben Abagyan, Chris de Graaf, Tracy Handel, Adriaan IJzerman, So Iwata, Ken Jacobson, Jonathan Javitch, Brian Kobilka, and October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 191

The Seven Transmembrane Superfamily 191

Bryan Roth. Lastly, I would like to thank Richard Lerner, Peter Schultz, Ian Wilson, and Jim Wells for being so consistently supportive of high risk/high reward research after more than 20 years of eﬀort.

References 1. R. Henderson, P. N. Unwin, Nature 257, 28 (1975). 2. K. Palczewski, T. Kumasaka, T. Hori, C. A. Behnke, H. Motoshima et al., Science 289, 739 (2000). 3. V. Cherezov, D. M. Rosenbaum, M. A. Hanson, S. G. Rasmussen, F. S. Thian distribution is allowed. distribution

et al., Science 318, 1258 (2007).

4. S. G. Rasmussen, H. J. Choi, D. M. Rosenbaum, T. S. Kobilka, F. S. Thian et al., Nature 450, 383 (2007). 9180 5. D. M. Rosenbaum, V. Cherezov, M. A. Hanson, S. G. Rasmussen, F. S. Thian

only. No further No only. et al., Science 318, 1266 (2007).

6. V. Katritch, V. Cherezov, R. C. Stevens, Annu. Rev. Pharm. Toxicol. 53, 531 (2013). 7. S. G. Rasmussen, B. T. DeVree, Y. Zou, A. C. Kruse, K. Y. Chung et al., Nature 477, 549 (2011). vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS 8. W. Liu, E. Chun, A. A. Thompson, P. Chubukov, F. Xu et al., Science 337,

- 232 (2012). 9. V. P. Jaakola, M. T. Griﬃth, M. A. Hanson, V. Cherezov, E. Y. Chien et al.,

http://www.sol Science 322, 1211 (2008). 10. R. C. Stevens, V. Cherezov, V. Katritch, R. Abagyan, P. Kuhn et al., Nat. Rev.

host on on host Drug Discov. 12, 25 (2013). 11. S. Knapp, P. Arruda, J. Blagg, S. Burley, D. H. Drewry et al., Nat. Chem. Biol. 9, 3 (2013). 12. M. Michino, E. Abola, J. S. Dixon, J. Moult, R. C. Stevens et al., Nat. Rev. Drug Discov. 8, 455 (2009). http://www.worldscientific.com/worldscibooks/10.1142/ 13. I. Kufareva, M. Rueda, V. Katritch, R. C. Stevens, R. Abagyan, Structure 19, 1108 (2011). 14. V. Katritch, V. Cherezov, R. C. Stevens, Trends Pharmacol. Sci. 33, 17 (2012). 15. M. A. Hanson, C. B. Roth, E. Jo, M. T. Griﬃth, F. L. Scott et al., Science 335, 851 (2012). 16. J. J. Liu, R. Horst, V. Katritch, R. C. Stevens, K. W¨uthrich et al., Science

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 335, 1106 (2012). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 192

192

NANOBODIES FOR THE STRUCTURAL AND FUNCTIONAL INVESTIGATION OF GPCR TRANSMEMBRANE SIGNALING

ELS PARDON and JAN STEYAERT Structural Biology Brussels, Vrije Universiteit Brussel, Pleinlaan 2, 1050, Brussel and Structural Biology Research Centre, VIB, Pleinlaan 2, 1050, Brussel distribution is allowed. distribution

My view of the present state of research on GPCRs and 9180 transporters: ligands, cofactors, drug development only. No further No only. Many membrane proteins including GPCRs and transporters are conformationally complex molecules. These are exiting times because we are starting to collect high resolution structures of the key functional conformers for a number of model systems. Most important, we are also witnessing the development of appropriate bio-

vayinstitutes.be/ physical methods to study the (thermo)dynamics of the conformational transitions PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- and the way ligands, cofactors, lipids or other proteins disturb these conformational equilibria. Ultimately, we will have to incorporate the contribution of the electro- chemical membrane potentials to fully understand transmembrane transport and http://www.sol signaling. No doubt that the remarkable progress that we are making in these ﬁelds will lead to better drugs. host on on host

My recent research contributions to GPCRs and transporters: ligands, cofactors, drug development

http://www.worldscientific.com/worldscibooks/10.1142/ Polytopic membrane proteins such as GPCRs and transporters are dynamic proteins that exist in an ensemble of functionally distinct conformational states. Crystal- logenesis typically traps the most stable low energy states, making it challenging to obtain agonist bound active-state X-ray structures of GPCRs. Stabilization of an active conformation of a GPCR can be achieved in diﬀerent ways. The most physiologic approach is to use a native signaling partner such as a G protein. An alternative to using a G protein is to identify another binding protein that can

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World stabilize the same conformational state. Antibodies evolved to bind to a diverse array of protein structures with high NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY aﬃnity and speciﬁcity. Last years, we generated Nanobodies (Nbs) that selectively recognize the active states of the β2 adrenergic receptor (β2AR) [1] and muscarinic

receptor 2 (M2R) [2]. Such Nbs that faithfully mimic the effects of the G proteins Gs or Gi were used to obtain diffraction quality crystals and to solve the very first structures of the active agonist-bound states of β2AR or M2R. More interesting, we also identified nanobodies that stabilize the β2AR·Gs complex [3]. One of these Nbs October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 193

Nanobodies for the Investigation of GPCR Transmembrane Signaling 193

was used to obtain the high-resolution crystal structure of this complex, providing the ﬁrst molecular view of transmembrane signaling by a GPCR. Our work illustrates the power of the Nanobody platform for GPCR research

[4]. Nanobodies are the small (15 kDa) and stable single domain fragments har-

boring the full antigen-binding capacity of the original heavy chain only antibodies that naturally occur in Camelids. Because of their unique three-dimensional structure, nanobodies have access to cavities or clefts on the surface of proteins. Our Nanobody discovery platform has the competitive advantage over other recombinant crystallization chaperones that the cloned Nanobody library represents the full collection of the naturally circulating humoral antigen-binding repertoire of heavy distribution is allowed. distribution chain antibodies, contrary to combinatorial libraries of conventional antibody frag-

ments. Because Nbs are encoded by single exons, the full antigen-binding capacity of in vivo matured antibodies can be cloned and eﬃciently screened for high aﬃn- 9180 ity binders, allowing one to fully exploit the humoral response of large mammals only. No further No only. against native antigens. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host

http://www.worldscientific.com/worldscibooks/10.1142/

Fig. 1. Nanobody-assisted X-ray crystallography of GPCR active states. (Left) Agonist bound active state of β2AR (ribbon representation) stabilized by Nb80 (surface representation); (Mid- dle) Active state of M2R (ribbon representation) bound to an agonist and a positive allosteric modulator that is stabilized by Nb9-8 (surface representation); (Right) β2AR-Gs transmembrane

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World signaling complex (ribbon representation) stabilized by Nb35 (surface representation). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Outlook to future developments of research on GPCRs and transporters: ligands, cofactors, drug development

Last years, we demonstrated that Nanobodies are exquisite tools for the structural investigation of conformationally complex membrane proteins. Future work will focus on Nanobodies that stabilize diﬀerent GPCR conformational states including October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 194

194 E. Pardon & J. Steyaert

active, inactive and biased conformations of the receptors, aiming at understanding how diﬀerent ligands elicit other biological responses. Our work on the β2AR-Gs complex indicates that nanobodies can also be used to stabilize protein complexes.

In the coming years we will develop methods to select nanobodies that stabilize tran- ... sient protein-protein interactions and focus on GPCR-G protein (Gs, Gi, Go, ), GPCR-arrestin and GPCR-kinase complexes and other signaling complexes involving β-arrestin, ERK2 and C-Src amongst others.

Acknowledgments distribution is allowed. distribution

The Steyaert lab was supported by the Fonds Wetenschappelijk Onderzoek-

Vlaanderen through research grants G011110N and G049512N, Innoviris Brussels through Impulse Life Science program BRGEOZ132, the Belgian Federal Science 9180 Policy Oﬃce through IAP7-40 and by SBO program IWT120026 from the Flemish

only. No further No only. Agency for Innovation by Science and Technology.

References 1. S. G. Rasmussen, H. J. Choi, J. J. Fung, E. Pardon, P. Casarosa et al., Nature vayinstitutes.be/ 469, 175 (2011). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 2. A. C. Kruse, A. M. Ring, A. Manglik, J. Hu, K. Hu et al., Nature 504, 101 (2013).

http://www.sol 3. S. G. Rasmussen, B. T. DeVree, Y. Zou, A. C. Kruse, K. Y. Chung et al., Nature 477, 549 (2011).

host on on host 4. J. Steyaert, B. K. Kobilka, Curr. Opin. Struct. Biol. 21, 567 (2011). http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 195

195

THE HIDDEN PHARMACOLOGY OF THE HUMAN GPCR-OME

BRYAN L. ROTH Department of Pharmacology and Division of Chemical Biology and Medicinal Chemistry University of North Carolina Chapel Hill Medical School, 4072 Genetic Medicine Building Chapel Hill, NC 27599, USA distribution is allowed. distribution

My view of the present state of research on GPCR drug discovery 9180 With nearly 900 members, the human GPCR-ome comprises the largest family of only. No further No only.

druggable targets in the genome although most members have not been chemically interrogated. G protein-coupled receptors (GPCRs) constitute the single largest class of druggable targets in the human genome [1]. The human genome includes genes ∼ vayinstitutes.be/ for 900 olfactory and non-olfactory GPCRs and careful analyses estimate the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS number of non-olfactory GPCRs to be between 342 and 356 (http://www.iuphar- - db.org/index.jsp) [2]. Of these, approximately 38% are classiﬁed as “orphans”, i.e., their natural (endogenous) ligands remain unknown (see IUPHAR GPCR Database

http://www.sol at: http://www.iuphar-db.org/index.jsp). Additionally a signiﬁcant proportion of the non-orphan GPCRs remain incompletely or inaccurately characterized with re-

host on on host spect to the ligands that modulate their activity.

My recent contributions to the area of GPCR drug discovery Over the past several years, my colleagues and I have shown that approved medica- http://www.worldscientific.com/worldscibooks/10.1142/ tions and their metabolites exert both known and unknown effects via unanticipated interactions with GPCRs [3–5]. In some cases, such as the infamous ‘fen/phen’ diet drug fiasco, an unanticipated interaction of a drug metabolite (e.g., norfenfluramine) with a GPCR (e.g.,the5-HT2B serotonin receptor) led to tremendous both human suffering and disasterous economic consequences [6]. In recent work done in close collaboration with Ray Stevens’ group we were able to provide a molecular explana-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World tion for how drugs can interact with these receptors and induce serious side-eﬀects [7]. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Most recently my group has been interested in using cheminformatics and other in silico approaches to facilitate GPCR drug discovery and genome-wide technolo-

gies to chemically interrogate the GPCR-ome and these will be discussed separately. Identifying ligands for orphan and non-orphan GPCRs Considerable eﬀort has been devoted to ﬁnding both endogenous and synthetic ligands for non-orphan GPCRs. For instance, we provided the initial validation data October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 196

196 B. L. Roth

[8] and later explored the utility of the Similarity Ensemble Approach (SEA; [8]) for identifying GPCRs as unanticipated targets of approved medications and chemical probes [5, 9–11] — all of these studies in collaboration with Brian Shoichet’s group at

UCSF. We have also used in silico approaches for drug design and lead optimization.

The most recent example of this was the successful design of multi-target drugs for aminergic GPCRs [12] using Baysean-based approach. In general, we have found these particular approaches to be useful for predicting small molecule interactions with molecular targets for which there is a large amount of annotation as to small molecule interactions. Unfortunately, for much of the GPCR-ome, little chemical matter is annotated (see Fig. 1). distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

Fig. 1. Annotated small molecules for the GPCR-ome. Shown is a scatter-gram listing of rep- http://www.sol resentative members of the GPCR-ome on the X-axis and # of small molecules annotated from Chembl. As can be seen, most of the GPCR-ome has relatively sparse annotation. host on on host

My lab has had some success with identifying ligands for orphan receptors [13–15] and this remains an active area of investigation.

http://www.worldscientific.com/worldscibooks/10.1142/ Genome-wide technologies to chemically interrogate the GPCR-ome Orphan GPCRs represent a particularly challenging target class because conventional GPCR screening approaches require either aprioriknowledge of the particular G protein(s) to which the orphan GPCR preferentially couples or the availability of positive control compounds (either agonists or antagonists) Various means to circumvent these challenges to provide so-called ‘universal’ GPCR screening platforms include the use of chimeric (or universal) G proteins, label-free impedance-based

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World approaches, arrestin-based screening platforms, melanosome-based approaches and others (see [16] for review), although these have not been applied on a large scale NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY to identify ligands for orphan GPCRs. We have been able to recently adapt an arrestin-based platform initially for D2-

dopamine receptor screening [17] and later for other receptors including k-opioid [18], serotonergic [7] and now nearly every GPCR in the human genome (see on- line resource at: http://pdsp.med.unc.edu/PDSP%20Protocols%20II%202013-03- 28.pdf; Kroeze et al., manuscript in preparation). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 197

The Hidden Pharmacology of the Human GPCR-Ome 197

Because the bulk of the GPCR-ome is not annotated from a chemistry perspective, developing genome-wide platforms to interrogate the GPCR-ome will greatly accelerate our ability to develop drugs and small molecule probes for the ‘hidden

GPCR-ome’. Additionally, as new chemical annotation will accelerate in silico

approaches such information will allow us to reveal the subterranean (or hidden) pharmacology of known drugs and bioactive compounds. Revealing such information will have a tremendous impact on not only drug discovery but also safety pharmacology and, ultimately, the health and well-being of humanity.

distribution is allowed. distribution Outlook to future developments of research on GPCR drug

discovery and chemical biology

Illuminating the pharmacology of the hidden GPCR-ome will transform drug dis- 9180 covery, although this will require novel and highly scalable technologies. only. No further No only.

Acknowledgments The work described here was supported by grants from the NIH (RO1MH61887;

vayinstitutes.be/ DA017204; an NIDA EUREKA Award; U19MH82441 and the NIMH Psychoactive PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

Drug Screening Program) -

References http://www.sol 1. A. L. Hopkins, C. R. Groom, Nat. Rev. Drug Discov. 1, 727 (2002).

host on on host 2. R. Fredriksson, H. B. Schioth, Mol. Pharmacol. 67, 1414 (2005). 3. R. B. Rothman, M. H. Baumann, J. E. Savage, L. Rauser, A. McBride et al., Circulation 102, 2836 (2000). 4. B. L. Roth, D. J. Sheﬄer, W. K. Kroeze, Nat. Rev. Drug Discov. 3, 353 (2004).

http://www.worldscientific.com/worldscibooks/10.1142/ 5. M. J. Keiser, V. Setola, J. J. Irwin, C. Laggner, A. I. Abbas et al., Nature 462, 175 (2009). 6. B. L. Roth, N. Engl. J. Med. 356, 6 (2007). 7. D. Wacker, C. Wang, V. Katritch, G. W. Han, X. P. Huang et al., Science 340, 615 (2013). 8. M. J. Keiser, B. L. Roth, B. N. Armbruster, P. Ernsberger, J. J. Irwin et al., Nat. Biotechnol. 25, 197 (2007).

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 9. E. Gregori-Puigjan´e,V.Setola,J.Hert,B.A. Crews, J. J. Irwin et al., Proc. Natl. Acad. Sci. USA 109, 11178 (2012). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 10. P. N. Yadav, A. I. Abbas, M. S. Farrell, V. Setola, N. Sciaky et al., Neuropsy- chopharmacol. 36, 638 (2011).

11. C. Laggner, D. Kogel, V. Setola, A. Tolia, H. Lin et al., Nat. Chem. Biol. 8, 144 (2011). 12. J. Besnard, G. F. Ruda, V. Setola, K. Abecassis, R. M. Rodriguiz et al., Nature 492, 215 (2012). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 198

198 B. L. Roth

13. B. L. Roth, S. C. Craigo, M. S. Choudhary, A. Uluer, F. J. Monsma Jr. et al., J. Pharmacol. Exp. Ther. 268, 1403 (1994). 14. T. Nguyen, D. A. Shapiro, S. R. George, V. Setola, D. K. Lee et al., Mol.

Pharmacol. 59, 427 (2001).

15. F. Oury, G. Sumara, O. Sumara, M. Ferron, H. Chang et al., Cell 144, 796 (2011). 16. J. A. Allen, B. L. Roth, Annu. Rev. Pharmacol. Toxicol. 51, 117 (2011). 17.J.A.Allen,J.M.Yost,V.Setola,X.Chen,M.F.Sassanoet al., Proc. Natl. Acad. Sci. USA 108, 18488 (2011). 18. H. Wu, D. Wacker, M. Mileni, V. Katritch, G. W. Han et al., Nature 485, 327 distribution is allowed. distribution (2012).

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 199

199

STRUCTURES AND REACTION MECHANISMS OF ABC TRANSPORTERS

KASPAR LOCHER Institute of Molecular Biology and Biophysics, Department of Biology, ETH Z¨urich; Schafmattstrasse 20, HPK D17, 8093 Z¨urich, Switzerland distribution is allowed. distribution

9180 My view of the present state of research on the structure and mechanism of ATP-binding cassette (ABC) transporters only. No further No only.

ABC transporters are primary active transporters that have an important role in the physiology and pathophysiology of many organisms. Fueled by hydrolysis of ATP, these proteins allow nutrients, metabolites, and noxious substances to be

vayinstitutes.be/ moved across lipid bilayers [1]. In bacteria, ABC importers facilitate the uptake PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- of nutrients from the environment via high-aﬃnity uptake pathways, thus enabling pathogenic bacteria to acquire essential (micro-)nutrients even when they are scarce. ABC exporters catalyze the extrusion of components required for the biosynthesis http://www.sol of the bacterial cell envelope and often contribute to drug resistance, which can pose problems in combatting pathogenic bacteria. In eukaryotes, ABC transporters host on on host are generally exporters that move diverse substrates across cellular or organelle membranes. Various hereditary human diseases have been associated with the dysfunction of ABC transporters, and their over-expression in tumor cells can cause multi-drug resistance, a serious impediment for the treatment of cancer [2].

http://www.worldscientific.com/worldscibooks/10.1142/ A great deal of progress has been made over the past 15 years in the structural and mechanistic investigation of ABC transporters. Specifically, X-ray crystallographic studies have revealed their folds and have visualized distinct conformations. The structural insight was useful for interpreting functional, biophysical, and biochemical studies and proved invaluable in understanding the coupling mechanisms of several model systems. While the first structures described ABC transporters of bacterial or archaeal origin, advances in over-expression and purification of eukary-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World otic membrane proteins have allowed the mechanisms of ABC transporters related

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY to human diseases to be tackled.

My recent research contributions to the structural and

mechanistic investigations of ATP-binding cassette transporters Research in my group at ETH Z¨urich has focused on the structure and mechanism of various membrane proteins, with a strong emphasis on ABC transporters. We October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 200

200 K. Locher

have spent considerable efforts at establishing and optimizing the over-expression and purification of ABC transporters in bacterial host cells, and more recently also in eukaryotic cells. Using X-ray crystallography, we have determined the first struc-

tures of type I and type II ABC importers, as well as that of the ﬁrst ABC exporter

[3–5]. The structures of these bacterial proteins were useful for studying the transport mechanisms not only of bacterial, but also of eukaryotic ABC transporters. Whereas the transmembrane domains of transporters belonging to diﬀerent subfamilies revealed distinct folds, the “motor domains” (nucleotide-binding domains) share a conserved fold and therefore also some similarity in binding and hydrolyzing ATP. We identiﬁed an essential structural motif termed “coupling helix”, which is distribution is allowed. distribution present in the transmembrane domains of all ABC transporters whose structures

were solved to date. This short α–helix is nestled into a groove formed at the sub- domain boundaries of the nucleotide-binding domains. The coupling helices provide 9180 the interface between the motor and the transmembrane domains, which is critical only. No further No only. for the transmission of conformational changes and thus for coupling ATP binding and hydrolysis to substrate transport (Fig. 1). vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host

http://www.worldscientific.com/worldscibooks/10.1142/

Fig. 1. Transmission interface in ABC transporters. The coupling helices are a conserved architectural motif of transmembrane domains. They are pushed together during ATP binding and pulled apart upon release of the hydrolysis products. This motion constitutes the power stroke of ABC transporters. The nucleotide binding domains form a closed, head-to-tail sandwich in the ATP-bound state. In the nucleotide-free state, they may either be spatially separated (lower left) or held together by C-terminal motifs or domains (upper left). ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World In 2007, by comparing distinct structures of transporters trapped in the ab- NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY sence of nucleotide or in the ATP-bound state, we deduced a coupling mechanism that could rationalize substrate uptake by type I ABC importers and extrusion

by ABC exporters [6]. The mechanism predicted that in the nucleotide-free state, ABC transporters adopt an inward-facing conformation, in which they release their substrates into the cytoplasm (importers) or recruit them from the cytoplasm or from the inner leaﬂet of the lipid bilayer (ABC exporters). In the ATP-bound October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 201

Structures and Reaction Mechanisms of ABC Transporters 201

conformation, the transporters adopt an outward-facing conformation, either re- leasing substrates (exporters) or acquiring them from the cognate binding proteins (importers). Alternating access was thus coupled to an ATP-dependent, motor

domain-driven reaction.

We have recently captured the vitamin B12 transporter BtuCD-F from E. coli (a type II ABC importer) in an intermediate state with bound AMPPNP, a non-hydrolysable ATP analog. Combined with functional data, this allowed a full transport cycle to be formulated. To our surprise, the structural rearrangements in the transmembrane domains of BtuCD-F are distinct from those of type I ABC importers or ABC exporters, suggesting that while the coupling helices distribution is allowed. distribution move in a similar way, the conformational rearrangements in the transmembrane

domains of ABC transporters diﬀer [7]. In BtuCD-F, two distinct cytoplasmic gates (compared to only one in other ABC systems) operate, allowing large substrates 9180 such as cobalamin (1355 Da in mass) to be transported without ion leakage to occur. only. No further No only.

Outlook to future developments in the structural Despite recent progress, the reaction mechanisms of ABC transporter–catalyzed

vayinstitutes.be/ processes are only understood for select substrates, all of which are water-soluble. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- For hydrophobic substrates, the transport mechanisms are poorly understood. This includes the transporters of toxic compounds, prescription or environmental drugs, or (glyco)lipid ﬂippases. Most of these reactions are highly relevant in pathophysiol-

http://www.sol ogy. For example, ABC transporter–catalyzed drug extrusion from endothelial cells of the blood-brain barrier protects the brain from toxic insult, but simultaneously

host on on host impedes the delivery of drugs for neurological diseases. The study of structures and mechanisms of ABC transporters of substrates that have a low hydrophilicity (drugs, lipids) will therefore likely unravel novel transport mechanisms. Such studies should have a potential value for future diagnostic or even therapeutic applications

http://www.worldscientific.com/worldscibooks/10.1142/ besides providing insight into elusive biological reactions.

Acknowledgments Research in my laboratory was generously funded by the Swiss National Science Foundation (SNF), NCCR Structural Biology, NCCR TransCure, Swiss Cancer League OncoSuisse, and ETH Z¨urich. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY References 1. I. B. Holland, S. P. C. Cole, K. Kuchler, C. F. Higgins, ABC Proteins: From

Bacteria to Man (Academic, London, 2003). 2. M. M. Gottesman, T. Fojo, S. E. Bates, Nat. Rev. Cancer 2, 48 (2002). 3. R. J. P. Dawson, K. P. Locher, Nature 443, 180 (2006). 4. K. Hollenstein, D. C. Frei, K. P. Locher, Nature 446, 213 (2007). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 202

202 K. Locher

5. R. N. Hvorup, B. A. Goetz, M. Niederer, K. Hollenstein, E. Perozo et al., Science 317, 1387 (2007). 6.R.J.P.Dawson,K.Hollenstein,K.P.Locher,Mol. Microbiol. 65, 250 (2007).

7. V. M. Korkhov, S. A. Mireku, K. P. Locher, Nature 490, 367 (2012).

distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 203

203

SESSION 4: GPCRs AND TRANSPORTERS: LIGANDS, COFACTORS, DRUG DEVELOPMENT

CHAIR: GUNNAR VON HEIJNE AUDITORS:C.GOVAERTS1,M.PARMENTIER2 (1) S.F.M.B., Universit´e Libre de Bruxelles, CP206/2, Campus Plaine, Bd du Triomphe, 1050 Brussels, Belgium distribution is allowed. distribution (2) WELBIO and IRIBHM, Universit´e Libre de Bruxelles, Campus Erasme,

808 route de Lennik, 1070 Brussels, Belgium 9180 only. No further No only.

Discussion among panel members Gunnar von Heijne: All right, thank you. That was very intense. I don’t think there’s a problem for the people in the panel to now come up with a load of questions to discuss but maybe I can just kick oﬀ by going back to one of the questions I put

vayinstitutes.be/ up on my initial slide which is — if we start with the GPCR ﬁeld — where would PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- you say that, you know, think about the basic mechanism of the GPCRs, what could you say that we kind of know for sure now about the mechanisms and where are we kind of speculating and hypothesizing? http://www.sol

Bryan Roth: We don’t know anything (laughs). So, I think we have some snapshots host on on host of some of the initial potential complexes but I think, as was mentioned, we really have no idea at the level of detail that we’d like to, about how the arrestin-GPCR complex is organized and how it functions. And then downstream from that there are, as Ray mentioned, potentially hundreds of other interacting proteins, which also

http://www.worldscientific.com/worldscibooks/10.1142/ are very important for signal transduction. I don’t think there are any complexes of those yet that have been reported with the GPCRs, so we’d like to see those from a structural perspective. And then, there’s this whole web of interactions inside the cell which is really a seamless web, which changes on a sort of millisecond basis and that is, I would say, completely unknown in terms of what is going on. So I would say, despite magniﬁcent progress in structural biology, we still are just at the very earliest stages of getting any sense of what’s really going on. So, I would say the

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Golden Age is yet to come. But we’re getting there. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Gebhard Schertler: So for me it’s quite exciting. You know, a few years ago, we were dreaming of seeing a ligand at all in a GPCR, and all we had were models, so

now we have really. . . every day we can see ligands, if we put the money in. So in terms of drug discovery cycle, we are really in a much better situation. If you want to try that, if we look at the understanding of the system, I showed you that we very well understand what the role of some of the conserved residues is, I think we also have seen there is a kind of general activation mechanism but there’s also October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 204

204 GPCRs and Transporters: Ligands, Cofactors, Drug Development

another very puzzling thing about GPCRs. They are actually very small, right? They’re tiny. But they interact with arrestin, G protein, can have diﬀerent signaling outputs. They will interact with the kinase, responding on the ligand you have and

then you will get a diﬀerent phosphorylation on the receptor. So there’s something

going on and so far the answer is that the different conformations that are possible, are the answer. But you know, let’s look at this a bit more complicated. If it’s chewing gum, then nothing is determined anymore, all right? So, how could it be a bit more systematically analyzed? So there are different answers. I showed you very conserved residues, so are there distinct states that we stabilize? Sometimes they have been called microswitches, so I think there are elements which are fixed but distribution is allowed. distribution because there are several of them they can be combinatorially switched, so I think in

this way there’s some underlying principle which allows us to give diﬀerent outputs. But it is not that every ligand induces a completely independent signaling pathway, 9180 it uses various sub-elements, and sub hydrogen-bond networks in the GPCR to only. No further No only. actually give you this new conformation. And this needs to still be proven, you know, we didn’t see so many states in detail that we really can see that, I mean. We have an impressive number of structures and sometimes it’s also a little bit hard to correlate the structures with what state it is. So we get the structure, we kind

vayinstitutes.be/ of look afterwards: What state it is? I have shown you a tryptophan on helix ﬁve, PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- which is sometimes in and out. That’s a little bit an indicator for at which point you’re in the activation cycle, or which state you have, but it’s actually not very strong and not 100% correlated but it be can used a little bit for that. So very http://www.sol often, after the structural biology, we imply what state it was. It’s not — even the complexes — it is not so clear what states exactly are relative to the other host on on host measurements. And I think this is a big problem. I think what would be very great is to know much more about the interactions with the kinases, there we have no structures, obviously we have a weaker interaction. Will it be really a full complex or is it a transient complex, can it be stabilized in the future? That would be very

http://www.worldscientific.com/worldscibooks/10.1142/ interesting. I think we will have very soon arrestin structures and we will see how they trigger actually other elements or the same element in the GPCR, but I think that will be resolved actually very, very soon.

Ray Stevens: I have a question for Brian, and this stems from yesterday’s discussions on the enzymes and stability. Gebhard just talked about stability. We don’t really characterize, we don’t consider stability to be a pharmacological property.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World But stability, how long a GPCR or a transporter resides in the membrane is incredibly important. We look at PK/PD eﬀects of drugs. Stability obviously is very NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY important for the actual protein targets. So why is receptor stability ignored?

Bryan Roth: I think because GPCRs are generally in terms of their residence in the membrane, and signaling, they’re not regulated by, so much by ubiquitination and intracellular proteolysis, that sort of things. It’s more a vesicular recycling mode of regulation. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 205

Session 4: Discussions 205

Ray Stevens: Well, that’s been our view. It’s all about traﬃcking, it’s about insertion, traﬃcking. But maybe it’s also about residence time.

Bryan Roth: Yeah, that’s an interesting idea and I don’t know what to say about that. I can sort of give you an anecdote from my past. So Anne Lesage is here as

an auditor and she was a post-doc with me when I was at Stanford. I had a failed postdoc there and was very lucky to get an academic position, but the reason the postdoc failed was that I thought that GPCRs were regulated transcriptionally and this was sort of early in the era when GPCRs were being cloned. And basically what I found out was that transcriptional regulation is not the way they are regulated and distribution is allowed. distribution

all the regulation apparently is post-translational and as far as is known, mainly,

basically turning the reactivity on and oﬀ, but not so much stabilizing their activity. Now it’s possible that there are things out there we don’t know about, related to 9180 GPCRs. But, I’ll keep an open mind. It’s possible that this is an area that has

only. No further No only. never been interrogated appropriately, but I’m not aware of anything related to that

with GPCRs.

Gebhard Schertler: So, another problem we have from the structural biology: we work on very strongly reduced systems. So we have removed N-termini, C-termini, vayinstitutes.be/ inserted loops, and we don’t think that they are biologically not important. Some PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- of them, have clear features from intrinsically disordered structures, like the C- terminus of rhodopsin. And I think in some GPCR situations they will be important

http://www.sol to assemble or co-assemble or bringing proximity, other factors, which are important in this signaling network. And we know very little about it, and actually we know

host on on host also not so well how to study this. All what people have done is: they have found hundreds of things that actually co-precipitate with GPCRs, so there are full lists of them. But there’s actually not a knowledge created from that. And this is potentially a great sea of new waves of pharmacological intervention. So because we are not knowing or understanding these signaling pathways and the signaling http://www.worldscientific.com/worldscibooks/10.1142/ routine, we really also can’t address this so well. Biased-signaling is great but, at the moment, it’s a molecular phenomenon. We are not linking it very well to human disease phenomenology, all right? So we don’t have fantastic examples where biased-signaling is clearly the cause of a side effect or clearly the cause of a cure, or clearly the cause of a disease. So it is much more complicated. However, finding this effect might explain certain unexplained things so far. We might think in another way about it. But we’re not immediately at this stage where we can say: we want ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World to bias something and then we get this medical outcome. So a lot to learn there. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY And it has to be partially done with patients or animal models and it cannot be done in cells or on the molecular side at all.

Bryan Roth: So there is one example of where biased signaling is unambiguously important. And this relates to the niacin ﬂush. I think I’m probably the only physi- cian here right? But maybe some of you have taken niacin for lowering cholesterol. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 206

206 GPCRs and Transporters: Ligands, Cofactors, Drug Development

It is very effective for lowering cholesterol and fortunately as effective actually as statins. The problem with niacin now is that many people when they take it they get a very severe flush, and it turns out that the flush is actually due to arrestin

signaling so Bob Lefkowitz and colleagues published a beautiful paper in the Jour-

nal of Clinical Investigation a few years ago where they were able to dissociate the flush side effect of niacin from the lipid lowering effects of niacin, and for actually what was previously an orphan GPCR. And so the idea, sort of going forward, has been to basically create a G protein-biased agonist for this particular GPCR, that clusters all lowering agents which don’t have this effect of flushing. So that’s one that’s key. There are a number of compounds in clinical trials, now from Travina, distribution is allowed. distribution which are G protein-biased compounds for the angiotensin receptor and we will

probably know about them shortly. And then, in full disclosure, my lab has a large collaboration with Pﬁzer pharmaceuticals to create arrestin-biased compounds for 9180 treating schizophrenia and those, at least in animal models are extraordinarily ef- only. No further No only. fective, and work only through arrestin. So I think these compounds certainly are percolating up and this is a really active area of drug discovery right now, and it won’t be too long I think before we have this compounds. That being said, it’s just one aspect of bias. So there all these others signaling pathways beyond arrestin

vayinstitutes.be/ that are not. . . PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Ray Stevens: So that was going to be my question to you, Brian. Everybody keeps talking about arrestin or G protein. Why is the ﬁeld so obsessed with those

http://www.sol two signaling pathways? It’s seems that it’s actually the name GPCR that has held back the ﬁeld because we have two hundred other intracellular proteins they

host on on host interact with.

Brian Roth: Yes, I agree 99.9 percent. The reason basically is for convenience that are, for those of you that have been involved in sort of lead optimization, things at the later stage of drug discovery need sort of very facile assays, and it’s http://www.worldscientific.com/worldscibooks/10.1142/ simply because the assays are available currently for G protein signaling and arrestin signaling and they are extraordinarily robust and selective and so. . .

Ray Stevens: Related to Al Edward’s comment — if you make a tool compound, all of a sudden everyone will start studying it. The appeal with G protein and arrestin signaling is the simplicity of the assay. It is what is available, and so that’s what everybody does. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Brian Roth: Right. Those are the assay, that’s what people do. We are trying to expand the universe, but yes there is no good reason other than those are the tools currently available that are in wide distribution.

Gebhard Schertler: I have a question for Kurt. First, I mean the NMR labeling studies are really fantastic because it directly showed that there are diﬀerent states, which we can observe in solution. What is your view of how NMR could look at the dynamics of GPCRs in more general terms in the next ﬁfteen years? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 207

Session 4: Discussions 207

Kurt Wüthrich: Well, as you know, NMR can do everything you want (laughs), as long as you have a sufficiently big magnet. To be serious, in β2AR we saw two states of two different sites in the protein, which means that we must have seen at least

three, possibly four states, depending on whether the inactive state is identical or

different in the two. And if we added more probes, the chances are that we would see additional states. So it would appear that we have a manifold of relatively long- lived states with locally different conformations. I did not mention this before: we have measured the lifetimes of the two states which we called A and I, and found a lower limit of about 1 second. So these are really long-lived states, meaning that the conformational change must involve major rearrangements, very likely including distribution is allowed. distribution backbone segments. You mentioned before that in order to get diffracting crystals,

you had to eliminate loops and truncate chain ends. In most cases, fusion proteins were used. Then you have to add CHS, and it is very critical which kind of detergent 9180 you use, and so on and so forth. And I mean, considering the amount of structural only. No further No only. information that has been obtained, it is amazing that you are not telling us how these GPCRs work. I’m surprised, you see, since we now have, at least, a hundred structures of 25 human GPCRs with diﬀerent ligands.

Ray Stevens: I’d actually say we’ve learned a tremendous amount. You’re right. So vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS right now there are more than two dozens diﬀerent types of GPCRs solved, there

- are over a hundred co-crystal structures. We have now started to decipher sodium effects and other solvent ion effects. Cholesterol is a big influence. We are beginning

http://www.sol to understand molecular recognition. So I disagree with your disappointment in the amount of what we’ve learned. host on on host Kurt W¨uthrich: Oh no no no, it wasn’t disappointment, it was surprise (laughs).

Ray Stevens: OK I’m surprised by your surprise (laughs). I think we’ve learned a tremendous amount. Now, we look to you and NMR, in terms of understanding http://www.worldscientific.com/worldscibooks/10.1142/ the dynamics, because we do think that these receptors, the way that they function is not from their static pictures. We think that it’s how they move and at what timescales, but I think we’ve learned a tremendous amount in the last ﬁve years.

Kurt W¨uthrich: Yes, we heard that one structure has been solved at room temperature.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Ray Stevens: That’s correct.

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Kurt W¨uthrich: You see, in solution NMR experiments we don’t have all these stabilizing interactions. We heard that a nanobody has to be added to stabilize

states of GPCRs that cannot otherwise be seen. Now, are these artifacts, or are these interesting states? It has also been pointed out that you have to bind the G protein or a nanobody to stabilize the activated form. We don’t have any of this, and in the absence of a cytoplasmic partner protein we get up to 90% population October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 208

208 GPCRs and Transporters: Ligands, Cofactors, Drug Development

of what we consider to be the activated state. So there are clearly data that need to be combined to sort things out.

Jan Steyaert: I think we all agree that GPCRs are highly cooperative things so that means that if you. . .

Kurt W¨uthrich: Highly?

Jan Steyaert: Highly cooperative molecules, so if you want to look at these highly cooperative systems, you cannot do it just with one, with two of the components. You at least need three components. You need a ligand at one end and you need distribution is allowed. distribution the cooperative partner at the other end. And so far we are always looking at the receptor with one of the partners, so to really understand how these things work we

9180 will need at least three molecular complexes: we will need a ligand at the outside and we will need the other partner at the inside. That’s the only way to look at only. No further No only. cooperative systems. And the GPCR by default is a cooperative system. So as long as you look at it from a two components system, I mean, you are missing, you always miss the downstream partner being it the G protein, being it the arrestin, so to really understand a cooperative system you need at least three components.

vayinstitutes.be/ And so far most of the structural information has been gained on two components PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- systems.

Kaspar Locher: Can I ask, Kurt, how do you measure your NMR spectra? What http://www.sol is the GPCR in? And is that a good model for a lipid membrane that contains up to 20% cholesterol? host on on host Kurt W¨uthrich: Well it contains 20% CHS. It’s a mixed micelle of 80% detergent and 20% CHS. It’s the same mixture that yielded crystals that diﬀract. Is that the answer to your question? http://www.worldscientific.com/worldscibooks/10.1142/ Kaspar Locher: So now the question is, if you’re asking the crystallographers or expressing your surprise that no clear consistent mechanism has come out and implied that NMR would solve that, my concern is that if you conduct these experiments at reasonably high temperatures I assume with 80% detergent, the dynamics could be altered and in ABC transporters a lot of dynamics are altered when you go into detergents so I guess there are question marks there too.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Kurt W¨uthrich: Yes, but I mean we don’t work at liquid nitrogen temperature. We

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY work at body temperature, not at “high temperature”. So which dynamics would be relevant?

Ray Stevens: You know, this has been a debate, and all of us in the room have at least heard of it this debate for forty years: NMR versus crystallography, solid state versus solution state. I heard Kurt talk 30 years ago against the evils of October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 209

Session 4: Discussions 209

crystallography, and I wish I would have recorded that lecture. I think the reality is, you use multiple techniques to try to understand what’s going on. We know that with crystallography, we are going to get a static snapshot, that’s the best that we

can do. But we get a pretty accurate static snapshot. Then we use NMR and the

dynamics, it’s the best that we can do. We would like to be able to do, in more sort of a lipid-like environment. We like using the lipid cubic phase because it’s a little bit more a lipid-like environment, but it’s all about combining the pieces together.

Gunnar von Heijne: Okay, so I think now it’s time for us to increase our coﬀee and cholesterol levels. So we will be back at 4.10 for the open discussion. Thank you. distribution is allowed. distribution

Discussion among all attendees

9180 Gunnar von Heijne: Right, who wants to start? only. No further No only. Kaspar Locher: I actually didn’t get a chance to ask a question at the panel discussion that I have of the GPCR fellows to the right of me. I had and remade this comment about not working with anything other than human, and I’m wondering how diﬀerent is mouse biology in terms of GPCRs from human biology? Because

vayinstitutes.be/ eventually, whatever drug you’re going to produce, and I think that’s one of your PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- goals, you’re going to produce a molecule that you going to have to put into a mouse, and I’m wondering what’s known about this.

http://www.sol Bryan Roth: So I preempt the question from Ray because we actually deal with this on a daily basis. In the first of the two serotonin structure papers, we go in host on on host this in great details. So there is this huge mouse/human difference in serotonin receptor pharmacology due to a single amino acid in the binding pocket, which actually killed drug discovery for Pfizer because they were developing compounds against the mouse receptor and it turned out that they didn’t interact with the

http://www.worldscientific.com/worldscibooks/10.1142/ human receptor. And this is sort of ubiquitous in GPCRs that there can be big species-specific differences in pharmacology in the binding pocket, usually due to a single amino acid differences. And so, as a practical matter, what’s normally done when you’re advancing compounds is to have the cognate receptor for every species that you’re looking at, for better or worse. So we have this project ongoing on serotonin receptors, which have the added complication that there are editing isoforms. So we have to have editing isoforms for human, mouse, and rat and every

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World compound that advances far enough has to actually be proﬁled in those before it

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY goes into the experimental animal. So it turns out for this particular family, if you want to be assured that what you’re looking at has a pharmacological relevance to the human, you have to have a human receptor, unfortunately.

Ray Stevens: I would add: each one of these structures is tough to solve, each one takes us anywhere from three to four years to solve. So, if you going to work that hard, you may as well go after the human. That’s sort of one. The second October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 210

210 GPCRs and Transporters: Ligands, Cofactors, Drug Development

is: it was mentioned that we use diﬀerent stabilization techniques. Chris Tate and Gebhard did beautiful work on stabilizing membrane proteins with mutations. We try to stabilize largely through chemical ligands. And again, most of the ligands

that have been developed are actually for human receptors, expressed in human cell

lines. Then from screening that way, eventually we get them into mouse in terms of studying some of the animal studies. Earlier in my career, we we’re working on two diﬀerent drug discovery programs, and we made great drugs that worked for mouse enzymes and it turned out they didn’t work all that well for human. And I decided, at the very beginning, we were going to focus on human. distribution is allowed. distribution

Gunnar von Heijne: But Kaspar, maybe we can throw the question back to you.

So in the transporter field, it’s mostly prokaryotic transporters so far. So how do you reason in this? 9180 Kaspar Locher: So, as I tried to point out, ten or twelve years ago when there only. No further No only. were no structures of ABC transporters it made perfect sense to use bacterial proteins as model systems to understand anything at all. Folds, coupling mechanisms and we’re finishing some of this work and looking at some of the reactions that are still poorly understood, like glycolipid flipping. Apart from that, I would agree vayinstitutes.be/ with Ray very much and the last two projects have shown how the human multidrug PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- transporters...Because it’s a very similar situation there. In mice, multidrug transport is a little diﬀerent and you do not want to have the situation where you’re

http://www.sol looking at pharmacological insight gained from structures only to find that it is different in the humans later on. So I agree very much. host on on host Ray Stevens: So, if I can ask one more question to Kaspar. You said that the transporter field is two decades behind the GPCR field. And I wonder why is that? Because, if I think about structural information, we saw a lot of transporters structures before we started see GPCRs structures. Is there something inherent http://www.worldscientific.com/worldscibooks/10.1142/ about the transporters? How are they different? Because everybody thinks their protein is the most unstable, most frustrating, annoying protein on the planet. And so what is it about transporters that makes them so challenging?

Kaspar Locher: The ABC transporters are the most frustrating (laughs). Just kidding. So we can’t talk about all transporters in one bag. There are many that are relatively easy to study. They are very diverse but I do think that their

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World importance for drug clearance, drugs moving around in the human body has been

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY underestimated. And this is catching up now, and part of the program that we have in Switzerland, which is called TransCure, is trying to remedy that, and looking at much more details of transporters involved in health processes. So, we’re behind in terms of interest and manpower of people who study this.

Richard Lerner: I want to raise an issue that Ray raised but went by very fast. And that is regulation by dwell time. The reason we became interested in that October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 211

Session 4: Discussions 211

is — you’ll see tomorrow just brieﬂy — we make agonist antibodies to receptors. And sometimes, you take a given cell clone, a stem cell let’s say, and you add a normal ligand, and it will make a white cell let’s say. And then you add the agonist

antibody and instead it will make a brain cell, and out of the same cells. And so

incidentally using the same downstream signaling pathways. So our model for this is a kinetic model, and that is that a dwell time keeps the signal on for a longer time and the cell can read signal length or signal gradient like happens in development biology or something like that. And the same receptor can be used in the same way depending on the dwell time. And I wonder if, in the case of the GPCRs in particular, what you term bias is a function of dwell time? distribution is allowed. distribution

Ray Stevens: So if I can make one comment to that Richard — on this whole issue of how long a receptor is sort of around. So I agree, the longer a receptor resides, 9180 the longer it can have an eﬀect on diﬀerent signaling pathways. We need to tease

only. No further No only. out what’s going on. But probably one of the most frustrating aspects at least of

the GPCR field is kinetics. You would think with these billion-dollar GPCR drugs available that pharmaceutical companies would have actually studied the kinetics of these different ligands, but if you want to know what the off-rate is of Plavix or some of these other drugs, it’s not known. In fact, pharmaceutical companies vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS do not study kinetics of receptor-ligands to the best of my knowledge. And we’ve

- begged, and we’ve asked for just about everything from the diﬀerent companies. So I think kinetics of binding is poorly understood, receptor residence time is very

http://www.sol poorly understood. We understand some aspects of traﬃcking absolutely. You know there is a really nice seminal work that has been done. C-terminus being a point

host on on host where the receptor gets pulled out of the membrane. What about GPCR kinetics?

Richard Lerner: But Ray, just to follow up with that, wouldn’t you think that as it were dwell time analoging would be an important, you know they’re more than happy to put a methyl on it. http://www.worldscientific.com/worldscibooks/10.1142/ Ray Stevens: Well if I look at PK/PD eﬀects of drugs, you would think that the drug companies and anybody that’s doing drug discovery, you would think people want to study the on- and oﬀ- rates of ligands as well, related to that, but they don’t. It’s hard. Part of the answer is that it’s just really hard to get this information and data and you want to do it on a large library of ligands, it’s even tougher. But I think it’s an important area of discovery. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Gebhard Schertler: I mean, it is interesting that G protein signaling normally is on the fast side, you know minutes to 20 minutes. And then, biased signaling, if you read it out, for example with kinetics, it does come later. So if you could have something diﬀerentiating and not being on long enough you might already get the bias. But it might not explain the whole thing. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 212

212 GPCRs and Transporters: Ligands, Cofactors, Drug Development

Kurt W¨uthrich: I’d like to come back to this monobody, two-body, three-body comment of before the break. In the monobody NMR experiments, with apo-β2AR, we have a visible population of the activated state with identical chemical shifts

to the states that we see with increased or decreased populations when diﬀerent

ligands are bound. This indicates that the diﬀerent states that we see are an inherent property of the GPCR and that it is not needed that a partner protein is associated with the GPCR to generate the activated conformation. I think, this is an important starting point for understanding how things happen. You see, you then have the receptor as a monobody, and you can start to add one, or multiple components and see how these components aﬀect the equilibrium between the two distribution is allowed. distribution conformations that preexist in the GPCR. I just want to emphasize this, when we

talk about mono, di- or tri-body systems, that we at least seem to have a starting point with a monobody. Of course, we do not know — now I can expand into 9180 the future — what the two states are that we see at each of these probe positions, only. No further No only. and we of course have to follow up with whatever method is available. We will try NMR structural work, that means isotope labeling of the GPCRs and deﬁning the structural variations between the two states that we see at each point. You see, with the ﬂuorine labels we have been able to measure the thermodynamics

vayinstitutes.be/ and the kinetics in the system of multiple pairs of two states, but we cannot make PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- statements about the structures behind it. For this, we need to collect additional data. That’s the outlook for the future.

http://www.sol Lode Wyns: I have a totally diﬀerent question, maybe linking this morning session and this afternoon. It’s well known there’s a number of GPCR-related diseases

host on on host where you have point mutations. It seems the GPCRs are perfectly happy, but signaling from the endosomes or signaling from the ER to a large extent seems to be OK but they are withheld and they don’t get to the surface. What is being done in this ﬁeld?

http://www.worldscientific.com/worldscibooks/10.1142/ Bryan Roth: So, there’s a whole area of pharmaco-chaperones, and I think there are actually a couple companies that have this. I think there are compounds that, so just to explain for people, there is this idea that when they’re misfolded, they’re stuck in the ER and there’s a concept called pharmaco-chaperones where typically there are actually drugs that are technically called inverse agonists. So they strongly stabilize the antagonist state of the receptor, and for many receptors, vasopressin receptor for which there are mutations that cause diabetes insipitus, which are due basically ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World to the receptor being trapped in the ER. So vasopressin receptor inverse agonist NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY actually then is able to make the receptor go to the cell surface. So, they’re out there. And I think, it’s you know, from a big pharma perspective it would probably

be a niche ﬁeld because they’re generally rare diseases but for those people that have them they are quite devastating and. . . So I think it’s moving along. I’m not aware of any drugs that are approved but I suspect that sometime in the future they will be out there. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 213

Session 4: Discussions 213

Ray Stevens: So, along those same lines, Brian happens to sit on the advisory board for the center that we have, and one of the things that we proposed last year was we want to start to study these splice variants and disease mutations. The advisory

board discouraged us from doing so because they said that the number was too

small. Do you remember that?

Bryan Roth: I remember that, yes. I think I said that.

Ray Stevens: For the GPCR family, they must mutate, but I think the bottom line is: it actually is rare. It doesn’t happen nearly as much as you would think. Now one of the best examples is the smoothened receptor. So Genentech has this great distribution is allowed. distribution

drug, GDC0449, for basal cell carcinoma. They give it to people, it does a great

job taking care of the cancer, but then months later the cancer comes back with a

9180 vengeance, massive. And, so what’s going on? And Genentech did some beautiful follow-up experiments. My understanding of the sort of results is that we all have only. No further No only. these variants in us, it’s just that the drug was hitting the wild type sequence, and those other variants were able to then come out and dominate, and so Lilly now has a follow-on drug. So I think there are some receptors, but smoothened again is not your canonical GPCR, it’s involved in embryonic development, which is a little vayinstitutes.be/ bit sort of unusual. So there are some examples but it seems to be fewer than I PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- personally would have thought, I thought it would have been many more.

Marina Rodnina: I have a question about optogenetics. Is it a useful tool for you http://www.sol in fact? And what are the perspectives of using that in pharmacology?

host on on host Ray Stevens: One of my graduate students just did his qualifying exam on optogenetics of GPCRs and it was a fantastic, he passed, so I was proud. I think the answer is: absolutely. One of the things we really want to try to understand is more about the sort of signaling, the traﬃcking, how these receptors are moving,

http://www.worldscientific.com/worldscibooks/10.1142/ how they are functioning in the cells. I think it’s a fantastic area of discovery. And it’s just starting to be pursued. I think Dennis was on the committee, any comments?

Dennis Wolan: Yeah, based on his proposal you mean, it was absolutely sound and very great potential to do these discoveries with this optogenetics.

Judith Klinman: I have a very naive question. From a mechanistic point of view, ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World where is the GTP actually hydrolyzed? Is it hydrolyzed? I mean these are G NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY coupled proteins, there is GTP getting hydrolyzed, no?

Ray Stevens: So G proteins are composed of three subunits, Gα, β and γ,andthe

enzyme reaction takes place in Gα. You have the receptor in the membrane, and then you have the Gαβγ complex, that is the structure that Jan showed from Brian Kobilka’s lab, that then forms a complex and then you get hydrolysis. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 214

214 GPCRs and Transporters: Ligands, Cofactors, Drug Development

Judith Klinman: OK, so for example Kurt in your experiment, where you’re looking at the two conformers, and you’re showing an interconversion between them with the agonist or the antagonist binding, if you put in a GTP analog that can’t get

hydrolyzed, can you start linking these conformational changes to chemical reaction?

Jan Steyaert: I think there’s a misunderstanding there. The structure is a nucleotide-free complex, and that’s exactly what the GPCR does, it just catalyzes, in enzymatic terms, the release of GDP, and then the complex falls apart. It binds GTP and the complex falls apart, and the hydrolysis happens in the free Gα later on, devoid of the G protein. distribution is allowed. distribution

Judith Klinman: I see, so he can’t even do the experiment. Okay, thank you.

9180 Gebhard Schertler: Yes but GTPγS is routinely used as G protein assay, you know. only. No further No only. Gunnar von Heijne: Other questions at this point? Before we leave the GPCRs, so maybe we will come back to them, I’m sure, but, you made a remark that was new to me and that was the suggestion that maybe GPCRs are channels, or transporters, some of them could be. So could you expand a bit on that? vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

Ray Stevens: When we started to get these high resolution structures, one of the - discoveries from 2011, was we saw this sodium binding site, and that was quite novel to us. We really didn’t expect it at all. And from that work then, we started

http://www.sol to do some follow-up experiments, looking at things that would dock in there, and we discovered amiloride, which is a channel blocker, that can actually bind into

host on on host that pocket quite nicely. So that started us thinking, and this is a hypothesis right now, that there’s enough space where sodium could actually pass through. It’s probably not enough space with some receptors to be a channel but perhaps as a transporter, it’s possible. So one of the areas that we are now investigating is

http://www.worldscientific.com/worldscibooks/10.1142/ trying to understand that possibility. We’ve also started looking at lithium, one of my big scientiﬁc curiosity is lithium treatment. How does it work? I asked Brian, he’s the sort of expert in this area, and he says “we really don’t know”. So we started looking at other ions, is there a possibility that lithium is going in this site, controlling the receptors? We ﬁnd rubidium is incredibly stabilizing and we think that rubidium goes into that pocket. Divalents don’t go in. So we are now starting the traditional soluble-protein type of studies now with the GPCRs to understand

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World this pocket and also if there is transporter or channel activity. The last comment on that is: every receptor is diﬀerent. We’ve now observed this in the delta opioid NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY receptor and in the adenosine A2A receptor. We think that this pocket is highly conserved in a lot of GPCRs. I think it’s now been seen by Gebhard, in the β1

adrenergic receptor. And so we’re starting to dig more and more into the exact role, and what role other ions might play with receptors.

Rudolf Glockshuber: I have a question to Kurt and all of you actually. So could your NMR experiments indicate that the drug binds according to a mechanism I October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 215

Session 4: Discussions 215

would say conformational selection. So, in the free state it exists in two diﬀerent conformations, and one of them is one that binds the drug. So the question is, are there other examples where one has something like an induced ﬁt mechanism, where

the structure of the drug-receptor complex is diﬀerent from any of the states that

are populated in the ligand-free state?

Gebhard Schertler: So the data from Kurt would look like that but there is actually something very curious about the GPCRs. If you put in, one of the very good synthetic agonist, isoprenaline into the receptor, the receptor does not go into an active conformation. So you do the structure of that, and this has been tried distribution is allowed. distribution

by Brian, and also with other agonists, you actually get a very ground-state-like

structure back. So the binding site for that is there but that doesn’t give you an open G protein binding site. So when you do ﬂuorescence lifetime measurement on 9180 a labeled receptor, again an experiment Brian did a very long time ago, and you

only. No further No only. put isoprenaline in, you can see again a few states, I don’t know if this has been

repeated with NMR, but what you see is that the active conformation is actually only increased in a tiny little amount. So to actually get activation of G protein and binding of the G protein and an exchange of nucleotide, you do not need all the receptor in an active conformation. And I don’t think this is a rare thing. So vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS there I don’t know exactly how this fits with induced fit. It might fit still in the

- ligand binding pocket. But it doesn’t give you immediately a complete ﬂip between two conformations. And that is why Brian has now crystallized with the G protein,

http://www.sol we have sometimes crystallized with the C-terminus of the G protein. There’s only one structure, and that is opsin — and we have also repeated that one — that

host on on host actually is there with a complete conformational change of helix 6 and that is the opsin structure that was observed at very low pH, in the Berlin group. So there’s some complexity also there. There is not a direct link between what you put in the ligand site, and the conformation on the intracellular side of the receptor.

http://www.worldscientific.com/worldscibooks/10.1142/ Ray Stevens: Back in 2007 when we published the β2 structure, one of our last sen- tences was: it’s going to be a long time before we have an agonist-bound structure, and it’s going to take G proteins. Now it turns out, that’s incorrect. People are getting agonist-bound structures left and right. And again, I’m not calling it an active state structure, I’m calling it an agonist-bound structure, but when we compared A2A, with the opsin — rhodopsin system, with the structure with G protein with β2, what we’re ﬁnding is lots of similarities. So, you can have an agonist bind, ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World and it preloads that binding site so. . . I’m just curious as to your thoughts. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Gebhard Schertler: No, I already said, you can bind the agonist but you don’t have ﬁxed movements of helix 6 for example, so there is not a direct linking between

these two sites. It looks more like a release of the helix.

Kurt W¨uthrich: Well, a direct answer to Rudi’s question is to say that the “principle of Le Chatelier” is acting here when we talk about binding of a drug molecule. The October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 216

216 GPCRs and Transporters: Ligands, Cofactors, Drug Development

drug molecule sees both conformations in the apo-protein. It binds to one, and removes it from the equilibrium, and Le Chatelier then pushes the equilibrium around. I think that’s the answer to your question as I would see it.

Rudolf Glockshuber: I think that’s quite clear. My question was rather are there

other examples where one has an induced ﬁt like mechanism, that maybe transmits to the cytoplasmic side.

Gebhard Schertler: So, when we go to larger ligands, there are examples where side chains are rotated. So for example I think in the muscarinic there’s one tryptophan that gets put out of the way so that a larger ligand can bind and then actually has distribution is allowed. distribution

a higher aﬃnity. So when we make our ligands in drug discovery bigger, we go out

of the orthosteric binding site of the natural ligand. Then some rearrangements are

9180 happening, at least of the side chains, and this has been observed in some places and also we have already seen an example in one structure where you had an allosteric only. No further No only. ligand bound, and we have kind of seen bivalent ligands which can actually reach both binding sites. So there are interesting detailed things coming about. . . In these cases, you have clearly also a change of the binding site, at least where you have one real rotamer change that’s induced by the ligand. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Tom Muir: So I think, Ray or Gebhard mentioned dimerization of the GPCRs. And now there’s a lot of back and forth on this. I’d just want to know what the current thinking was, from all the structures, based on crystal packing, you know, http://www.sol the Kobilka structure, would there be room for two G proteins within a dimer, would they occupy the same space. Where are we with that? host on on host

Ray Stevens: Bryan Roth helped organize a meeting at the NIH last summer. So we were just discussing this at the break. What happened in the ﬁeld was that a lot of people studied GPCR dimerization related to GPCR signaling and they

http://www.worldscientific.com/worldscibooks/10.1142/ used overexpression systems where they made lots of receptors, and I think the general attitude is that with many of that, much of that data is artifactual. It’s overexpression and you force dimerization to occur and you saw what you wanted to see. Now, that being said, I personally believe that there are some cases where there are true dimers. I think the class C receptors, the mGLUrs. I think that data is pretty solid where they are dimers. I think personally something like CXCR4, where we crystallized it in five different forms, different crystal forms, different ligands,

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World diﬀerent variants, and every time, it was always the same dimer. Now chemokines

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY CXCR4 are chemical gradient systems. Where it’s all about concentration of the receptor, and concentration of ligand, and you can have a 1:1, 2:1, 1:2 ratio. I think it’s receptor speciﬁc. I was just talking to Bryan Roth at the break where, I still like to fantasize that you can have some receptors that heterodimerize, and that make some sense on biology: dopamine with adenosine, but he says that he has data that says: that’s an artifact. So it’s an open question for the ﬁeld right now. It’s going October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 217

Session 4: Discussions 217

to be receptor speciﬁc. So people have to be very careful. One of the things that we’re learning, and I’m glad that we looked at more than one GPCR for the past 5 years, I’m glad that were going to look at all 826, because if you focus on one, and

you try then to make correlation to other receptors, you get tripped up. And so I

think it’s important to study the whole family.

Bryan Roth: So, let me just expand on this. This is something that my lab spent a lot of time unsuccessfully trying to replicate, and I won’t state what the papers are but some of the more high visibility heterodimers papers, we were never able to replicate the data. Despite, you know, expending vast resources to do so, multiple distribution is allowed. distribution

knock-out mice. One of these has actually been formally non-replicated by Abbott

labs. This was the 5hT2A-mGluR2 that was in Cell and Nature. I mean it could happen but it’s not robust enough to be replicated independently by at least one 9180 group. So it is, I would say sort of a contentious area, that you could sort of populate

only. No further No only. this group here with people that believe — I would say believe — in heterodimers

strongly, as well as people that don’t. So I think it’s at that level. It’s still, still a bit up in the air, and the thing that I always look for are the independent replications. I think this is the key. One of the nice things about the crystal structures, and I think the opioid receptors were really, really nice examples of this. All four came vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS out basically in the same issue by two independent groups, and they were very

- similar. And in one case the nociceptin receptor had a diﬀerent fusion partner, and a completely diﬀerent part of the receptor, and intact intracellular loops and, but

http://www.sol the overall packing, and everything was very, very similar. So, this is the sort of things that you like to see where multiple groups are essentially getting the same

host on on host data. You don’t really see this so much in the dimer field. It’s more one-off type things so I’m still waiting for the really strong signal from multiple independent labs replicating something. But I think it’s something, certainly something to be aware of. But my personal success is that we can’t replicate this stuff.

http://www.worldscientific.com/worldscibooks/10.1142/ Bernard Henrissat: How much of the incredibly detailed knowledge that you guys are accumulating can be transmitted to other species? You know, we don’t care whether some animals get sick, but just out of curiosity: if you go from mammalian systems to a non-mammalian system, if you go to insects, birds, if you go to plants, are these drugs working the same?

Gebhard Schertler: I think at the end we don’t know, there are very famous exam- ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World ples where it doesn’t work, like before I would like to mention the β3 project, which

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY was done for obesity. It was again done on rodents, and it didn’t work in humans, so even small diﬀerences can make a diﬀerence, but if I look just at the sequence space, and the conserved residues, and the mechanism I tried at least to sketch to

you with a movie, clearly all the residues are there, and the principal activation mechanisms are the same, but, you know the ligand sites are very finely tuned, and also the signaling network is very finely tuned, and you very easily get different outputs from that. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 218

218 GPCRs and Transporters: Ligands, Cofactors, Drug Development

Dario Neri: The title of this session covers also drug development. There were some aspects touched, for example by Gebhard where structural knowledge would facilitate for example fragment-based drug discovery. I would be curious if you

could expand a little bit on where and how structural knowledge helps drug devel-

opment and maybe also limitations, in translating structural information into drug development. Maybe one needs to close one cycle of. . .

Gebhard Schertler: The ﬁrst thing I want to say is obviously I was one of the people making Heptares, I have shares of it. In addition, I’m on the scientiﬁc advisory board; I’m not really allowed to tell you exactly what they do. What they have distribution is allowed. distribution

done is that they have solved many in-house structures. They really have solved

ﬁrst structures with no ligands, then with ligands they made and then also with fragments in about that order. They have now also diﬀerent classes, and there was 9180 a publication of another class of receptors recently. What is very important is they

only. No further No only. are not using only the information from only one receptor. They have assembled a

fantastic modelling team. Every model that they use is derived from more than one structure, so every time we get a new fragment, a new structure, they will integrate that in their modelling eﬀort. It is actually a very strongly model-driven company and they have given quite a lot of talks also about the necessity to either know or vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS model the waters in the binding sites, and they have for example very much this

- idea that there are two classes of waters, some that are in good energetic positions, and you shouldn’t necessarily replace them by the ligand, others which you could

http://www.sol well replace. A lot of this is really very high end modelling, you could believe it or not but there have been reasonably successful with this hypothesis. So they have

host on on host ﬁlled the pipeline, and they have a lot of partnership deals, but I think they also have very skilled experts doing this, it’s not only the structural information that is pushing this. They have exceptional chemists, they are very fast, you know, always doing a series, and not one compound. So drugs and receptor structure is one driver of drug discovery, it’s not a drug discovery pipeline. I really don’t believe at all. http://www.worldscientific.com/worldscibooks/10.1142/

Ray Stevens: I will make the same disclosure. About the same time Heptares got started, I started a company called Receptos. Receptos is four years old, and it has a S1P1 molecule in phase III, and did its IPO last May. Structure was not used — the molecules came from Hugh Rosen and Ed Roberts with med-chem design. In this case, structure was helpful in understanding how diﬀerent molecules bound and what was unique And most importantly in that case, structure was used there were ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 763 diﬀerent ligands to look at. How do you prioritize those that go into clinical NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY trials? So structure was used as to try to prioritize things, and use that knowledge combined with med-chem intuition and everything else. So I think, clearly structure

is being very helpful. Now, the other point is the use of virtual ligand screening in drug discovery to find new molecules. If I look at kinases and enzymes, we have hit rates of 5%. If you find a hundred different molecules by virtual ligand screening, you’re lucky if five of them then get validated. But with GPCRs it’s very, very October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 219

Session 4: Discussions 219

diﬀerent. It’s a hit rate of 40, 60, 80%. So the question is: why? We’ve done this with Dopamine D3, we’ve done it with A2A, Brian Shoichet has done it with some nice systems. I think the reason is, you have these beautiful three-dimensional

pockets, unlike enzymes. They are actually more rigid than we thought they were

going to be, and because you get these beautiful three-dimensional pockets, to then do your docking, we’re seeing great validation. Brian said it earlier, the best is yet to come. You’re going to see a gold mine in the structures that are just coming out now. So you’ll see a lot of structure-based drugs emerging over the next couple of years. distribution is allowed. distribution

Bryan Roth: So, let me expand on this. Let me just say I have nothing to disclose,

because I don’t own a company. But what we’ve done is mainly published in collaboration with Brian Shoichet, and my own lab. It’s something we use every 9180 day, basically, and if you have a structure, obviously, it makes things way better.

only. No further No only. And it greatly accelerates the pace of things. But it’s a tool. That being said, the

last slide I showed was the discovery of this positive allosteric modulator for this orphan receptor. The thing that may not have been clear from this slide is: that was discovered in silico, entirely in silico. So this is a receptor for which there is no structure. It was made completely based on homology modelling, and we did vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS some mutagenesis with the very low aﬃnity compound, and then based on docking

- with the Shoichet group we were able to this extremely potent compound which is actually eﬀective in vivo. So, it’s possible to go all the way with docking. I

http://www.sol think this is probably going to be a rare case, but it does work. So it certainly is a fantastic tool that we have now, that we just didn’t have before. host on on host Dennis Wolan: So GPCRs have fantastic ligand eﬃciency, very well known for that. So what kind of promiscuity is there across these molecules? I mean you have 826 GPCRs, and you have assays for majority of these?

http://www.worldscientific.com/worldscibooks/10.1142/ Bryan Roth: Yes.

Dennis Wolan: So, if you have assays for the majority, are they fairly similar assays?

Bryan Roth: So I didn’t really get into that cause I didn’t want to get into that degree of granularity, but it’s basically an arrestin-based assay, which is cross platform. So technically it allows us to screen all the “druggable” GPCRs in the genome in a

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World single 384-well plate. And it had sort of “pennies per well”- it’s highly scalable. So what we’ve done, I didn’t show these slides, we looked at the world’s most promis- NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY cuous compound which is clozapine, which before we had this platform we knew to bind to probably a hundred diﬀerent molecular targets, and then — I have a slide,

so if anybody is interested I can show it to them — then we screened it against the entire GPCR-ome, and what we found was that there were no other targets. Basically it binds to the biogenic amine sort of subfamily of receptors. It hits all of October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 220

220 GPCRs and Transporters: Ligands, Cofactors, Drug Development

those and then hit some other enzymes but outside of that, interestingly enough, it has no activity. And we screened both highly promiscuous kinase inhibitors and other promiscuous drugs, and you don’t sort of see “panGPCR” activity like you

would with the kinases, but you do see interesting hot spots. And in many cases

these, we can basically deconstruct the information back to a known side effect of the drug or a known therapeutic effect of the drug, that was previously uninterrupt- able. We were basically able to find a molecular target for that, but it’s not, unlike the kinases, or phosphatases or something like that, where we have an inhibitor of one, it hits everything. This does not appear to be the case outside the biogenic amine family. distribution is allowed. distribution

Ray Stevens: So I showed this phylogenetic tree based on sequence. Brian Shoichet has this really nice chemical version of that map. And if you look at for example 9180 muscarinic receptors and chemokines, there’s a lot of overlap there. Why that is?

only. No further No only. Nobody knows. But it’s one of the areas that’s pretty cool for discovery. We have

a table in the lab, of again a lot of ligands do bind multiple receptors, and there are a lot of good examples of it. And so it’s another area of investigation.

Kurt W¨uthrich: Eventually you want to ﬁnd correlations between the chemical vayinstitutes.be/ structure of the ligand, of the drug, and its consequences for what happens on the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- cytoplasmic end of the molecule. Now, if we had induced ﬁt as Rudolf Glockshuber inquired, then we would have very broad promiscuity. I think that’s perfectly clear.

http://www.sol If we had induced ﬁt, then the whole system would be highly promiscuous. On the other hand, as far as I can oversee the results of the crystallographic studies,

host on on host the changes in the global architecture, induced by agonists and antagonists, are very limited. When you go through the series of now 25 GPCRs whose structures are here, the major differences are in the ligand-binding site. And that’s what I understand, these are the different sizes of your wine glasses, and the different shapes of your wine glasses in your slides. If we had induced fit, this would all be http://www.worldscientific.com/worldscibooks/10.1142/ essentially of limited use.

Gebhard Schertler: Why do you think then, all these residues are conserved if they play no role?

Kurt W¨uthrich: I don’t understand your comment.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Gebhard Schertler: I showed you, at least three motifs which are dispersed from the binding site towards the intracellular side, and they are conserved in the family A NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY receptors and I also showed you that they build two alternating hydrogen networks, and they are very heavily conserved in the chemokine receptors as in rhodopsin as

in amine receptors. And what you’re saying is, the binding site, they play no role. I haven’t said that, in principle. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 221

Session 4: Discussions 221

Kurt W¨uthrich: I think this is a misunderstanding. I understand that there are major variations in the ligand binding site, whereas in the scaﬀold of the GPCRs there are only rather small changes.

Gebhard Schertler: I agree on that.

Jan Steyaert: Can I comment on this?

Kurt Wüthrich: So this would indicate that specificity is due to the mean structures seen in the crystal structure determinations, whereas an induced fit mechanism would lead to random promiscuity. distribution is allowed. distribution

Jan Steyaert: Can I comment on that, because if we compare the active structures

9180 against the inactive structures, so that meaning, a site occupied by an antagonist or an inverse agonist and compared that to the same receptor loaded with an agonist, only. No further No only. you clearly see a compacting, a smaller binding pocket, and you can even observe, in the absence of the G protein. . . Well, let’s turn it around: in the presence of the G protein, the oﬀ-rate of ligands really increases dramatically, so the ligand is really binding that small pocket that it’s made for, only in the presence of the G

vayinstitutes.be/ protein, and you see factors of 200 even 1000 in terms of slowing down oﬀ-rates, PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- and on-rates also slow down, so there’s an aspect of induced ﬁt in these binding pockets, upon binding of agonist, natural ligands, and the concomitant binding of the G protein. http://www.sol Ray Stevens: Some of this is actually semantics because, again, if you look at

host on on host adenosine A2A, we now have, and I think probably close to twenty co-crystals structures, diﬀerent ligands, from everything, part- full agonist, partial antagonist and so forth. And it’s been remarkable how similar those structures are. One of the things that’s been nice is: it’s been solved with antibodies, with fusions, with

http://www.worldscientific.com/worldscibooks/10.1142/ mutations, and the structures are quite similar. The reason I say semantics is those binding sites, within each specific receptor, the differences are not as large as I think a lot of us thought they were going to be. This is all perspective. Myself I was expecting the agonist structure to be very different from the inverse agonist. And it’s not nearly as big as I expected. So again; perspective. And if we look at all the different receptors we’re seeing a sort of similar trends. The off-rate thing I think is a fascination. I mean, how on earth is it having this effect on off-rates, and changes

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World are pretty small. In the top half we have ligand recognition — very small changes. The bottom half is where you’re seeing the bigger conformational changes, and how NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY that’s working, and it’s quite interesting how it can be few residues making that happen. So I agree with Kurt.

Gebhard Schertler: Can I go in there? I really disagree here, because we have two structures which have bits of G protein bound, we have Brian’s, and we have the rhodopsin one, with at least the C-terminal end. And in my lab, we’re also do- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 222

222 GPCRs and Transporters: Ligands, Cofactors, Drug Development

ing experiments on the complex. So the ﬁrst thing you observe, if you make the rhodopsin complex, again with the empty pocket to G protein, that actually the retinal after hydrolysis cannot leave anymore. Okay, so there is a strong change,

but that’s not on ligand binding, that’s when the G protein has bound. And some

of the structures you’re describing are without the G protein but the agonist bound. They are very similar, and they are all still in a similar conformation as the ground state structure. They have not made that big transition. But when you do that big transition there’s no question from the literature that the aﬃnities change dramatically when you add the G protein, and also they change when you add the peptide on it, the C-terminal peptide of the G protein. At that moment, when you really distribution is allowed. distribution bloom out the cytoplasmic side, you have changes it at the extracellular side, and

you’re starting to trap the ligand. And if you look carefully at the structures with the antibody and with the G protein complex, it is very clear that the open cleft 9180 in the β receptor has closed up, and actually an agonist now cannot come in and only. No further No only. replace the ligand anymore. These are absolutely clear measurements, there’s no question about this.

Ray Stevens: I think the diﬀerence of opinion is: what deﬁnes a big conformational change? I’m thinking 8 or 10 A˚ in the direct binding pocket. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Gebhard Schertler: I think we are talking about diﬀerent states in the cycle of doing this. You’re talking about an early one, which I’ve called at some point an

http://www.sol encounter complex when the agonist is going there, and what we’re here talking about is the one that has bound the G protein and at that point we’ve actually a

host on on host strong enough conformational change that even the extracellular side is starting to close up. And this you only can see when you have the G protein or a G protein substitute there.

Ray Stevens: Well let’s see when we get more structures. http://www.worldscientific.com/worldscibooks/10.1142/ Markus Grütter: I would like to switch to the transporters and ask a question to Kaspar. You have shown in one nice example, the importer, how substrate is transported into the cell. Now there’s a whole, large number of ABC transporters that export and that are promiscuous in the way they transport substrates, and we have seen structures of different conformations but not really a convincing site where things bind. How do you see the future with respect of seeing how stuff is

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World transported through this? Apart, I mean, we see the mechanistic movements, but

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY where are the substrates? Do you see a chance to tackle that experimentally and how?

Kaspar Locher: Maybe I should explain ﬁrst for the non-experts that one of the dif- ﬁculties of working with ABC exporters is that most of their substrates are so hydrophobic that they compete very strongly, or that detergent competes very strongly for the binding pocket. So whenever we purify them, they are devoid October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 223

Session 4: Discussions 223

of any substrate, and the substrate don’t bind well anymore, and we have to put the transporters back into a lipid bilayers, cholesterol-containing lipid bilayers often, to see the substrate even binding and then usually with lousy activity of tens of

μmol. So unlike in the GPCR ﬁeld, there’s a very diﬀerent mechanism of substrate

recognition. And the current ideas are that somewhere on the molecule, and I don’t think anybody has a clear idea of where that is, there are multiple residues, could be hydrophobic, aromatic, that contribute to drug binding. How that works and how it’s extracting a drug from the favorable interactions that the drug experiences in the lipid bilayer is unclear. But what is clear is that side openings to the lipid bilayer play an important role. How large those are again is a matter of debate. distribution is allowed. distribution And my personal view is that whenever we take a multidrug or any other ABC ex-

porter out of the membrane, with detergents, it will relax into a non-physiological conformation that’s the most stable conformation, that’s why we can also crystalize 9180 it. And the openings, and the distances of NBDs are as large or larger than those only. No further No only. where we can prove inactivation of the transporter. So no clear answers in the near future, there are some systems with good inhibitors and we’re currently working with chemists to improve those inhibitors and hopefully get to higher aﬃnities that will allow us to get a structural insight even in detergent solution. Is that the answer

vayinstitutes.be/ to your question? PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Markus Gr¨utter: Yes, and how about assays to tackle these questions?

http://www.sol Kaspar Locher: Right, so we approach the idea of assays as follows: until now, assays have mostly looked at the stimulation of ATPase activity. So the idea being, we add

host on on host a drug into a reconstituted ABC transporter-containing vesicle, it partitions into the lipid bilayer and it binds to a drug binding pocket which is assumed to be a central drug binding pocket thereby speeding up ATPase activity. The problem with this assay is that even DMSO and other solvents will have the same effect. And you can affect ABC transporters even by adding, by changing the lipid composition. What I http://www.worldscientific.com/worldscibooks/10.1142/ think we will have to do and what we’re working on is to develop very good acceptor proteins that will prevent partitioning of the substrate into the lipid bilayer, and really allow the substrates to be kept outside of the lipid bilayer. After all that’s exactly what Nature does. So if you look at the blood-brain barrier, the reason why many of the drugs that we would like to give for brain diseases don’t enter the brain is the blood-brain barrier, and there are multiple multidrug transporters expressed there and the reason they work is blood flow and albumin. So the drugs would ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World be transported out of the membrane, they would instantly partition back into the NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY membrane unless there was blood flow and albumin. And similar things we have to do in assays.

Markus Gr¨utter: Thank you.

Gebhard Schertler: Maybe one comment to Kaspar. When we look at rhodopsin in the opsin form, and this is work from Peter Hofmann, he saw really a nice channel October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 224

224 GPCRs and Transporters: Ligands, Cofactors, Drug Development

for the hydrophobic retinal going in sideways into the G protein-coupled receptor a little bit like you suggest in the transporter. But it’s only in one conformation. These openings are really closed in the ground state-like structure.

Kaspar Locher: Right and perhaps I would like to, I didn’t get a chance to go to

the channel, the potential channel features of a GPCR. I would assume this would be measurable, and I can only assume at the moment that these rates are so slow that they wouldn’t be above leak currents of empty membranes. So I doubt very much that GPCRs are transporters of any signal cations.

Ray Stevens: Can we bet on that? distribution is allowed. distribution

Kaspar Locher: I don’t have your purse but I’d be willing to bet (laughs). 9180 Ray Stevens: I’d like to bet beer. But I have sort of insider information so. . . only. No further No only.

Bryan Roth: So, I would not take that bet, Ray.

Ray Stevens: I have a question for Gunnar. So we’ve been talking about GPCRs and transporters. What about, in terms of their insertion into the membrane. I vayinstitutes.be/ mean, what is it going to take for us to understand how these receptors, eukaryotic PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- proteins are inserting and folding/refolding.

Gunnar von Heijne: I don’t think anything is known about it at this point. Obvi- http://www.sol ously they do get in, that’s about as much as we can say I think. They do get in co-translationally in, I think that’s clear as well. And they use the SEC translocon host on on host but beyond that I don’t think we can say anything really. I think we have looked at a few of those just theoretically to look at how hydrophobic these transmembrane segments are. And some are not super hydrophobic. So, there’s, you know, although nobody else looked at it, I think the bet is again — if we should start a http://www.worldscientific.com/worldscibooks/10.1142/ betting club here — that there’s going to be interactions, co-translational interactions between at least some of the helices as they assemble into the membrane. How much of the final fold that’s going to be co-translational versus post-translational I don’t think anybody can say. But the first elements, I mean we’ve seen this with another protein, multi-spanning protein, that there is in fact early co-translational interactions between some helices or polar residues on some helices that can help stabilize not so hydrophobic helices across the membrane. But there are very few ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World assays around to look at it, and you know even with purified reconstituted mem-

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY brane proteins it’s very diﬃcult to do unfolding-refolding studies. There are two or three systems where you can do that, basically reversibly and that’s of course a far cry from co-translational feeding of transmembrane helices into membranes.

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 225

Session 4: Discussions 225

Markus Aebi: Following up on that, nevertheless there’s a quality, or let’s say a quality control system for proper folding of membrane proteins. And I wonder how this is recognized by this system.

Gunnar von Heijne: That’s, as far as I know, that’s a post-translational quality

control system. So, I mean, clearly many of these proteins don’t attain their full three dimensional structure until fairly late. Whether there are chaperones involved, there are a few cases, where it has been shown that there are intramembranous chaperones. The cases I know of, they are very speciﬁc, so there is like one chaperone for one protein, but nothing excludes that there could be more general chaperones distribution is allowed. distribution

for instance that we just haven’t come across yet. And so, you know, exactly what

these states are that are recognized as misfolded we also don’t really know. We know that they behave as, in some sense where they can be recognized by the cell 9180 as being misfolded, but exactly what they look like is not clear I think. only. No further No only. Kaspar Locher: You have various systems where you can hold the translocation halfway. And in eukaryotes, we heard about the OST complex and it’s assumed that there’s a direct association with the secretion machinery. Is there an example where a halted peptide that’s only semi inserted into the membrane already get vayinstitutes.be/ glycosylated? PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Gunnar von Heijne: Oh yes, I mean you can make truncated ribosome nascent chain complexes where, which have been targeted to the translocon and the nascent http://www.sol peptide is halfway across the membrane and some part is in the lumen, some part is spanning the translocon and some part is stuck in the ribosome. And those can host on on host be glycosylated on the luminal side. There is a minimal distance away from the membrane which is basically I think reﬂecting the distance between the membrane and the active site of the OST in that case. We’ve been trying to do, to see if we can do the same in bacteria. . . You know we just don’t have a great experiment http://www.worldscientific.com/worldscibooks/10.1142/ yet but I think we can see a similar kind of thing, but that’s with the substrate, that’s not folded, so it’s you know quite unfolded in that case.

Dennis Wolan: So, speaking glycosylation, Ray, with your 1.7 A˚ structure that you have, are you seeing glycosylation, and is there any biological signiﬁcance?

Ray Stevens: We give these things haircuts, they are glycosylated when they’re

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World made, we use PNGase F and we clip oﬀ as much as we can, but we see beautiful sugar stems from them. So that’s the answer to your question. I want to sort NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY of follow up, with Gunnar, I mean, one of the things we know is: these receptors are glycosylated. You know that if you knock out the glycosylation you don’t get

expression to the surface. So you know the glycosylation is important. We also know that we can express different parts. I think Brian Kobilka who did the experiment years ago where he co-expressed I think it was the first 4 helices and the last 3 helices, maybe it was 5 + 2 of the different helices. With better knowledge and October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 226

226 GPCRs and Transporters: Ligands, Cofactors, Drug Development

questions, we’re now repeating the experiment of 4 + 3, where we know, the four helices, the ﬁrst four helices form this beautiful four-helix bundle. And we’re trying to see if we can crystallize just that. It doesn’t have a function, but it’s a curiosity.

And can we start to combine these things together and restore function to try to

get this question of how are these things assembling? So is it partial folding?

Gunnar von Heijne: Yes of course those types of experiments have been done with many multi-spanning membrane proteins, starting with bacteriorhodopsin, way, way back. So they can be. . . Not all combinations work apparently, and I don’t think anybody has ﬁgured out why certain combinations work and others don’t, whether distribution is allowed. distribution

it has to do with insertion or something else but they do seem to work. In many

cases they can ﬁnd each other in the membrane somehow.

9180 Kurt W¨uthrich: Well, earlier today, there was the question of the lifetime of membrane proteins in living systems, and this of course is connected with folding and only. No further No only. insertion. What is known about the turnover of membrane proteins, of GPCRs, of channels?

Bryan Roth: Yeah, so, I know this. There’s a system that we developed, may be

vayinstitutes.be/ or may not be familiar to you, where we use engineered GPCRs to regulate the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- signaling called DREADDs, and a colleague of mine at Scripps, Mark Mayford did the experiment. So they’re expressed using a tetracycline inducible promoter, so he can turn on expression on the brain with TET, and then take it oﬀ, and they’re http://www.sol gone in two days. So the turnover is actually relatively quick, surprisingly, so this is a basically a muscarinic receptor, sort of prototypical family A GPCR. So at host on on host least for that receptor we know quite well. Whether that will be the same for other GPCRs — God only knows — but you know it was actually fairly quick, I wouldn’t have suspected that it would be gone so quickly after basically the transcription was shut oﬀ. And that’s in the absence of ligand. So it’s not being activated by http://www.worldscientific.com/worldscibooks/10.1142/ any endogenous ligand.

Gunnar von Heijne: So maybe one interesting example of differential stability but within a complex is of course the photosynthetic reaction center, right? You get photo-damage to the central unit, subunit, and that can be picked out of the complex specifically and replaced with a fresh one, without affecting the other subunits. So even within one complex, you know, you can have different stabilities. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Kurt Wüthrich: Would that mean that their individual domains fold independently? NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY And then assemble as a whole domain?

Gunnar von Heijne: Yes, there’s some work done on the assembly of these multicomponent complexes. They do follow not completely deﬁned but partially deﬁned assembly pathways where, l say, subunits A and B have to assemble before subunit C can come in, but maybe subunit D could come in before and after subunit C so October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 227

Session 4: Discussions 227

it’s not straight linear. But it has certain structure. In the cases that we’ve looked at — at least these more complex multicomponent membrane proteins systems.

N.B. Some of Professor Stevens’ contributions to this section’s discussion were added

in the proofs. distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

July 25, 2013 17:28 WSPC - Proceedings Trim Size: 9.75in x 6.5in icmp12-master

This page intentionally left blank distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 229

Session 5 Biologicals and Biosimilars distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 230

distribution is allowed. distribution

Modeled structure of the PASylated Fab fragment of an antibody. See

9180 Figure 2 contributed by Arne Skerra on page 246. only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 231

231

BIOLOGICALS AND BIOSIMILARS

SRABONI GHOSE and MARKUS G. GRUTTER¨ Institute of Biochemistry, University of Z¨urich Winterthurerstrasse 190, CH-8057 Z¨urich, Switzerland distribution is allowed. distribution

Introduction 9180 Biologicals are drugs produced using biotechnology. They may be proteins, includ- only. No further No only.

ing antibodies, or nucleic acids including DNA, RNA or antisense oligonucleotides, that are produced using recombinant technology as opposed to extraction from existing sources e.g., from human blood. Biotechnology has made a wide range of new therapies possible by allowing the large scale production of vaccines, therapeu-

vayinstitutes.be/ tic protein hormones, cytokines, tissue growth factors, monoclonal antibodies and PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

oligonucleotides for cell or gene therapies. In this discussion we will focus on the - categories of biomolecular drugs (biologicals) that are presently managed by the FDA Center for Drugs Evaluation and Research (CDER): monoclonal antibodies,

http://www.sol cytokines, tissue growth factors and therapeutic proteins. The ﬁrst genetically engineered biomolecular drug to receive US approval was human insulin in 1982. Since

host on on host then, blood clotting factors, growth hormone, erythropoietin, colony stimulating factors, interferon and anticoagulant proteins, all natural proteins or ‘ﬁrst generation’ recombinant products, have been brought to the market. Likewise monoclonal antibodies have entered the market. Biologicals today represent a major business

http://www.worldscientific.com/worldscibooks/10.1142/ segment in the bio/pharmaceutical industry. Based on worldwide sales, seven out of the top 15 pharmaceuticals in 2012 were biologicals [1a]. The therapeutic potential of biologicals is evident in the growing number of product approvals by the FDA and the EMA, 39 in the period 2009–2011 [1b]. Cancer and related conditions remain the main indication. As the ﬁrst generation biologicals start losing patent protection, and biosimilars enter the market, innovation is seen in the design of next generation drugs as well as in the biomanufacturing process. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY First generation biologicals Two broad categories of protein biologicals are discussed here:

(i) Proteins with an enzymatic or regulatory activity which (a) replace a deﬁcient protein such as insulin, (b) augment an existing pathway such as erythropoietin and (c) modulate an activity such as botulinum toxin October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 232

232 S. Ghose & M. G. Gr¨utter

(ii) Proteins with a highly speciﬁc targeting activity which (a) directly interfere with an activity such as Trastuzumab (Herceptin) or (b) deliver a therapeutic compound such as a radioligand (131I-tositumomab).

Antibody technology Besides the ongoing search for high affinity full length therapeutic monoclonal antibodies from libraries, the field has matured through the generation of murine to chimaeric to fully human antibodies. Antibodies in addition are further modified and engineered for higher affinity and stability. Antibody fragments such as distribution is allowed. distribution Fab, scFv and Fc have improved diffusion profiles due to their smaller molecular

weight, thus offering distinct advantages over full length antibodies [2]. Bi-specific T-cell engaging antibodies, Fc domain libraries, dual affinity re-targetting, are all 9180 engineered antibody fragment technologies now in clinical trials for different indica- only. No further No only. tions. Antibody-drug conjugates (ADC) now offer great potential for development especially since linker engineering has increased stability in the bloodstream and efficient payload release at the target site. As of 2013, 30 ADCs are in clinical trials targeting 24 antigens in oncology indications [3]. Peptibodies are biologically active

vayinstitutes.be/ peptide-Fc fusions that allow for increased aﬃnity as well as eﬃcient manufacturing PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- [4].

http://www.sol Oligonucleotide technology The ﬁrst systemically delivered antisense drug mipomersen was approved for the

host on on host treatment of homozygous familial hypercholesterolemia (hoFH) in 2013. The early promise of oligonucleotide therapeutics was announced with the approval of the ﬁrst drug in 1992 against genital warts. Drug delivery remains a signiﬁcant hurdle. RNA therapeutics are particularly suited to restore gene function in condi-

http://www.worldscientific.com/worldscibooks/10.1142/ tions where blocking oligonucleotides do not degrade pre-mRNA [5, 6]. Recent encouraging results from clinical trials for Duchenne muscular dystrophy and inherited neurodegenerative diseases are a further example of its potential. Besides RNAi and antisense oligonucleotides, splice-switching and translation-suppressing oligonucleotides as well as siRNA, are being developed for therapeutics.

Biomanufacturing ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World The manufacturing of biologicals has had to mature as new processes were estab- NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY lished on an industrial scale. Developments such as protein engineering, synthetic gene design, recombinant expression technology and large scale fermentation and

extraction have all contributed to the scaling-up process. The criteria for biomanufacturing are very diﬀerent from small molecule manufacturing and new regulatory guidelines had to be deﬁned to guarantee quality standards. The objectives include limiting the immunogenicity of the biological and improving its shelf life. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 233

Biologicals and Biosimilars 233

Research tools have been developed to characterize the biologicals with regard to their aggregation status, three-dimensional structure and post-translational modi- ﬁcations [7]. The majority of biologicals are glycoproteins, the eﬀect of the glycan

composition and its engineering has implications for product quality [8]. It has been

seen that not only does the quality of the product to be guaranteed but also the quality of the process, thus Quality by Design is now also applied to biomanufacturing.

Biosafety and pharmacovigilance

distribution is allowed. distribution The safety proﬁle of biologicals is diﬀerent from small molecule drugs due to their

inherent characteristics conferred by the complex composition. These large, some-

times heterogeneous mixtures of isoforms have a greater potential for immunogenic- 9180 ity. These drugs are often used to treat chronic conditions and thus the risk factor is increased. The spontaneous reporting of adverse drug reactions is vital to evaluating only. No further No only.

the safety of these drugs in the clinic. An example of dramatic and unpredictable consequences is the cytokine storm that occurred with the anti CD-28 monoclonal antibody (TeGenero) study, which was not observed in preclinical studies. The clinical trial design of biologicals has been reviewed after this unfortunate episode. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Biosimilars Biosimilars are defined as biotherapeutic products which are similar in terms of http://www.sol quality, safety and efficacy to already licensed reference biotherapeutic products. Thus, the biosimilar has the same pharmacological effect without the same chemi- host on on host cal identity as the patented drug. It has been argued that biosimilars of monoclonal antibodies will be difficult to generate since each one is unique and that the different process will result in a different product. However, biosimilars have been approved several times both by the FDA and the EMA. The first monoclonal an-

http://www.worldscientific.com/worldscibooks/10.1142/ tibody biosimilar was approved in the EU in 2013. The use of biosimilars is set to dramatically increase as the first generation biologicals lose their exclusivity [9]. The clinical studies carried out so far have been positive. New paradigms for testing are required for the comparability exercise of complex drugs such as Avastin as it loses exclusivity in 2017 in the USA. About 20 different biosimilars of Herceptin are now in development. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Second generation biologicals — novel protein scaffolds NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Although monoclonal antibodies are very effective therapeutic agents, certain inherent characteristics limit their broader use in the clinic. The large size

(150–160 kDa), the poor tissue penetration in case of solid tumours and the structure which does not easily allow binding to enzyme pockets are undesirable features [10]. The next generation biologicals are selected high-affinity binding proteins from rationally designed scaffold protein libraries which offer distinct advantages over a October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 234

234 S. Ghose & M. G. Gr¨utter

full length antibody. These binders have already been very useful tools in research, in the crystallization of membrane proteins for example and as antibodies. Display technologies have greatly contributed to the generation of these designed libraries.

The ﬁrst scaﬀold protein biological to be approved was ecallantide, a designed Ku-

nitz domain inhibitor of human plasma kallikrein.

Nanobodies Nanobodies are single domain antibodies with a molecular weight of 12–15 kDa [11]. Derived from the variable VHH domain of the llama, they have to be humanized distribution is allowed. distribution to reduce the immunogenicity. Several nanobodies are in clinical trials for diﬀerent

indications such as rheumatoid arthritis and osteoporosis. 9180 Scaﬀold proteins only. No further No only.

DARPins DARPins are designed ankyrin repeat proteins with a molecular weight of 14 or 18 kDa depending on whether 4 or 5 repeats are present [12a, 12b]. Well expressed

vayinstitutes.be/ in E. coli, DARPins are heat stable and may be highly concentrated. DARPins have PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- been successfully used to target tumours and in diagnostic applications. Several DARPins are in clinical trials in ocular indications. http://www.sol Anticalins

host on on host Anticalins are a class of proteins derived from a human lipocalin scaﬀold with a molecular weight of about 20 kDa [13]. Lipocalins are abundant in plasma and tissue ﬂuids and well suited for clinical applications due to their low immunogenicity. One anticalin is in an early stage clinical trial in an oncology indication. http://www.worldscientific.com/worldscibooks/10.1142/

Affibodies Affibodies are high affinity antibody mimetics derived from the IgG binding domain of S. aureus Protein A with a molecular weight of only 6 kDa. The scaffold is stable at extreme pH and high temperature [14]. Radiolabelled affibody-albumin conjugates have been used for diagnostic molecular imaging in oncology. Affibodies

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World are also in clinical trials in an oncology indication. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Monobodies/adnectin

Adnectins are extracellular type III domain of human ﬁbronectin derived scaﬀold proteins with a molecular weight of about 10 kDa [15]. The domain has a structure similar to antibody variable domains. An adnectin is in a phase II clinical trial as an anti-angiogenic inhibitor in an oncology indication. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 235

Biologicals and Biosimilars 235

Domain antibodies A domain antibody (dAb) is either the variable domain of an antibody heavy chain (VH domain) or the variable domain of an antibody light chain (VL domain) [16].

Dual targeting human domain antibodies are now in a Phase II clinical trial in a neurology indication.

Fynomers, Nanofitins, Knottins, Avimers, Affilins Fynomers, derived from the novel, fully human protein scaffold Fyn SH3 domain, have a molecular weight of under 10 kDa [17]. Nanofitins are structurally derived distribution is allowed. distribution from the DNA binding protein Sac7d, found in Sulfolobus acidocaldarius,anarchae-

bacterium [18]. They are unusually heat resistant scaffolds with a molecular weight of about 7 kDa. Knottins are engineered cysteine knot peptides for high affinity 9180 molecular recognition against tumor-associated receptors [19]. Avimers are evolved only. No further No only. from human LDL receptor domains by in vitro exon shuffling and phage display, generating multidomain proteins with binding and inhibitory properties [20]. Affil- ins are structurally derived from human γ-B crystallin (molecular weight about 20 kDa) or ubiquitin (molecular weight 10 kDa) [21]. All these scaffold proteins are in

vayinstitutes.be/ the preclinical phase of drug discovery. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Outlook

http://www.sol This brief overview shows the extent to which biologicals have developed in the last 3 decades. Antibody engineering has given rise to a new set of binders with innovative

host on on host characteristics such as bi-functionality and conjugation to chemo- or radiotherapic loads. The future looks bright as the second generation pipeline has an impressive collection of drugs designed to overcome the hurdles of delivery, stability and access to the antigen. Biosimilars have entered the market as the ﬁrst generation biologicals lose their exclusivity and, having won approval, will be a big player also driving http://www.worldscientific.com/worldscibooks/10.1142/ innovation in biomanufacturing. The long term eﬀects of these drugs have to be closely monitored, requiring clear regulatory and pharmacovigilance guidelines.

Acknowledgments Funding from the Swiss National Science Foundation, NCCR Structural Biology and the University of Z¨urich is gratefully acknowledged. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY References 1. (a) http://www.ﬁercepharma.com/special-reports/15-best-selling-drugs-2012

(b) http://www.biopharminternational.com/biopharm/article/ articleDetail.jsp?id=775167 2. S. J. Kim, Y. Park, H. J. Hong, Mol. Cells 20, 17 (2005). 3. A. Mullard, Nat. Rev. Drug Discov. 12, 329 (2013). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 236

236 S. Ghose & M. G. Gr¨utter

4. G. Shimamoto, C. Gegg, T. Boone, C. Qu´eva, MAbs. 4, 586 (2012). 5. R. Kole, A. R. Krainer, S. Altman, Nat. Rev. Drug Discov. 11, 125 (2012). 6. A. L. Southwell, N. H. Skotte, C. F. Bennett, M. R. Hayden, Trends Mol. Med.

18, 634 (2012).

7. S. A. Berkowitz, J. R. Engen, J. R. Mazzeo, G. B. Jones, Nat. Rev. Drug Discov. 11, 527 (2012). 8. N. Lingg, P. Zhang, Z. Song, M. Bardor, Biotechnol. J. 7, 1462 (2012). 9. M. Rinaudo-Gaujous, S. Paul, E. D. Tedesco, C. Genin, X. Roblin et al., Ali- ment Pharmacol. Th. 38, 914 (2013). 10. T. Wurch, A. Pierr´e, S. Depil, Trends Biotechnol. 30, 575 (2012). distribution is allowed. distribution 11. F. Van Bockstaele, J. B. Holz, H. Revets, Curr. Opin. Investig. Drugs 10, 1212

(2009). 12. (a) R. Tamaskovic, M. Simon, N. Stefan, M. Schwill, A. Pl¨uckthun, Meth. 9180 Enzymol. 503, 101 (2012). only. No further No only. (b) H. K. Binz, P. Amstutz, A. Kohl, M. T. Stumpp, C. Briand et al., Nat. Biotechnol. 22, 575 (2004). 13. M. Gebauer, A. Skerra, Meth. Enzymol. 503, 157 (2012). 14. S. Hoppmann, Z. Miao, S. Liu, H. Liu, G. Ren et al., Bioconjug. Chem. 22, 413

vayinstitutes.be/ (2011). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 15. M. Ackermann, B. A. Morse, V. Delventhal, I. M. Carvajal, M. A. Konerding, Angiogenesis 15, 685 (2012). 16.L.J.Holt,C.Herring,L.S.Jespers,B.P.Woolven,I.M.Tomlinson,Trends http://www.sol Biotechnol. 21, 484 (2003). 17. D. Grabulovski, M. Kaspar, D. Neri, J. Biol. Chem. 282, 3196 (2007). host on on host 18. B. Mouratou, G. B´ehar, L. Paillard-Laurance, S. Colinet, F. Pecorari, Meth. Mol. Biol. 805, 315 (2012). 19. S. J. Moore, J. R. Cochran, Meth. Enzymol. 503, 223 (2012). 20. J. Silverman, Q. Liu, A. Bakker, W. To, A. Duguay et al., Nat. Biotechnol. 23,

http://www.worldscientific.com/worldscibooks/10.1142/ 1556 (2005). 21. H. Ebersbach, E. Fiedler, T. Scheuermann, M. Fiedler, M. T. Stubbs et al., J. Mol. Biol. 372, 172 (2007). ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 237

237

PLATFORM TECHNOLOGIES FOR THE ARTIFICIAL PSEUDO-NATURAL PRODUCT DISCOVERY

HIROAKI SUGA Department of Chemistry, Graduate School of Science, The University of Tokyo 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan distribution is allowed. distribution

My view of the present state of research on biologicals and

9180 biosimilars only. No further No only.

Peptide-based molecules are appealing scaffolds for the development of novel drugs. Their high structural diversity, often combined with a large surface area for interaction with the target, allows for very high affinity to the target with exceptional specificity, as evidenced by the success of peptide therapeutics derived from natu-

vayinstitutes.be/ rally occurring peptide hormones [1–3]. Moreover, even relatively small peptides are PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

able to bind to target sites that are not the active sites of enzymes and to modulate - protein-protein interactions, characteristics which are not generally exhibited by traditional small molecule drugs. However, despite these characteristics, peptides

http://www.sol exhibit some features that are undesirable in potential therapeutics. Most notably, their susceptibility to proteases makes them unstable in vivo, they are not generally

host on on host orally available, and they are unable to cross cell membranes making them unsuited for use against intracellular targets. Interestingly, a number of relatively small peptides (generally of the order of 15 residues or less) derived from natural sources circumvent these apparent limitations

http://www.worldscientific.com/worldscibooks/10.1142/ (Fig. 1). Perhaps the best example of this is the immunosuppressant cyclosporine, a highly N-methylated, macrocyclic, 11 residue peptide which is orally available, exhibits acceptable pharmacokinetics and exerts its therapeutic eﬀect by binding to intracellular target proteins, cyclophilins [4]. Other clinically useful peptide-based drugs derived from natural sources include the echinocandins (anti-fungal), actino- mycin D (antibiotic, but used clinically as a chemotherapeutic), and the antibiotics daptomycin and vancomycin [5]. Taken together, these examples demonstrate that

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World these relatively small naturally occurring peptides can exhibit “drug-like” properties, combining the pharmacokinetic qualities of small molecules with the target NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY binding features of larger peptides. Notably, almost all naturally occurring peptides with potent bioactivities and fa-

vorable pharmacokinetic properties possess non-canonical structural features. The most frequently observed of these is some form of macrocyclic motif, which confers structural rigidity (and thus increased target binding aﬃnity) and resistance to proteases, and also plays an important role in biological membrane permeability [6]. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 238

238 H. Suga

distribution is allowed. distribution

9180 only. No further No only.

Fig. 1. Flexizymes. Depending on the substrate kinds, three ﬂexizymes are available.

N-alkylation (commonly N-methylation) of the peptide backbone is also commonly

vayinstitutes.be/ observed, and has also been shown to be important for membrane permeability PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

(presumably by reducing the hydrophilicity of the amide backbone) and protease - resistance. Additionally, d-stereochemistry, non-canonical side chain structures, backbone heterocycles, and other non-standard moieties are frequently observed

http://www.sol in cyclic peptides derived from biological sources, and appear, like macrocyclization and N-alkylation, to imbue peptides with improved pharmacokinetic proﬁles

host on on host relative to canonical peptides. The appeal of macrocyclic peptides as drug candidates has lead to significant research efforts to develop screening methods capable of isolating peptides with high target binding affinities from complex libraries. Whilst combinatorial chemistry

http://www.worldscientific.com/worldscibooks/10.1142/ approaches have shown some promise, several more recently developed techniques have employed biosynthesis of macrocyclic peptides from libraries of combinatorial genetic templates. Such strategies allow for the synthesis of very diverse libraries and the use of iterative selection/ampliﬁcation-based screening techniques, so that libraries of macrocyclic peptides greater than 1013 species can be rapidly screened for binding to a target of interest. Furthermore, the application of genetic code manipulation techniques to these biosynthetic systems and/or post-translational

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World chemical reactions have greatly expanded the range of chemical moieties that can be synthesized, allowing for the screening of libraries of artiﬁcial macrocycles that NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY mimic the structural features of naturally occurring bioactive peptides, i.e.,artiﬁcial pseudo-natural products.

My research themes focus on the development of novel platform technologies that enable us to generate genetically encoded artiﬁcial libraries of macrocycles and pseudo-natural products, and the use of selection methods to isolate such molecules with not only high target aﬃnities but also drug-like properties. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 239

Platform Technologies for the Artiﬁcial Pseudo-Natural Product Discovery 239

My recent research contributions to biologicals and biosimilars Flexizymes and ﬂexible in-vitro translation (FIT) system: Tools for the genetic code reprogramming

Flexizymes are artiﬁcial ribozymes with highly promiscuous aminoacyl-transferase

activity. The discovery of these ribozymes through in vitro selection experiments, and details of their properties and use has been recently reviewed elsewhere [7]. In brief, ﬂexizymes are small (45 or 46 nucleotides depending on the kinds, Fig. 1) artiﬁcial RNAs that catalyze the aminoacylation of essentially any tRNA, using an amino acid substrate that has been chemically activated by a leaving group distribution is allowed. distribution

attached through an ester linkage [8]. Available three ﬂexizymes are capable of

charging virtually any amino acid (or any other small molecule with a carboxyl moiety that can be activated) onto diverse tRNAs bearing any anticodon desired. 9180 Importantly, the codon reassignments with non-canonical amino acids using

only. No further No only. ﬂexizymes are executed by pre-charged non-canonical aa-tRNAs. Therefore, the

vacant codons for reassignment can be created by omitting not only the relevant amino acids but also their cognate ARS. Since trace amounts of canonical amino acids are often carried into the reconstituted translation system with the puriﬁed ribosomes, translation factors and/or tRNAs, the double exclusion of amino acids vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS and cognate ARSs ensures complete knockdown of any potential competition for

- elongation. Moreover, aa-tRNA synthesis using ﬂexizymes is far more facile than the chemo-enzymatic method, and ﬂexizymes will accept a much wider range of

http://www.sol amino acid substrates than ARSs including N-alkyl-amino acids [9, 10], D-amino acids [11], N-acylated amino acids [12], and even exotic peptides containing β-and

host on on host γ-amino acids [13]. Introduction of the resulting non-canonical aa-tRNAs into the custom-made translation system in which speciﬁc codons have been vacated thus enables the incorporation of any translatable amino acid(s) into a polypeptide at any desired position(s). Such a translation system integrated with ﬂexizyme chem- istryisreferredtoasaflexible in vitro translation (FIT) reaction. Consequently, it http://www.worldscientific.com/worldscibooks/10.1142/ has been possible to generate extremely diverse chemical structures based loosely on a peptide backbone. These have included peptoids (N-alkylated polyglycines) [9], (poly)esters [14] and peptides with diverse patterns of N-methylation and non- canonical side chains [10]. Furthermore, the diverse chemistries accessible through the FIT system permit the synthesis of peptides containing reactive moieties that can mediate intramolec-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World ular cyclization. This allows for the synthesis of macrocyclic peptides without the need for intermolecular chemical cross-linking. As a speciﬁc example, the FIT NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY system can be used to synthesize a chloroacetylated N-terminus, which will spontaneously react with a downstream cysteine residue to form a macrocycle through

a non-reducible thioether bond [15]. Other examples of macrocyclization through introduction of reactive moieties by FIT synthesis have included oxidative coupling, Huisgen 1,3-dipolar cycloaddition, Michael addition, and native chemical ligation October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 240

240 H. Suga

strategies, and a number of these have been combined to produce bicyclic peptides with deﬁned structures [15].

RaPID system: A tool for the discovery of non-canonical

macrocyclic peptides The combination of the FIT system with an mRNA display-based screening method is known as random non-standard peptides integrated discovery (RaPID, Fig. 2), and has been used to identify a number of bioactive macrocyclic peptides with high binding aﬃnities to therapeutic targets of interest. Like the ARS-mediated genetically reprogrammed mRNA display selections described above, RaPID screening distribution is allowed. distribution

involves translation of a semi-randomized puromycin-linked mRNA library in a

genetically-reprogrammed, fully-reconstituted FIT system, such that the resulting

9180 non-standard peptides are covalently linked to the mRNA that encodes them. Itera- tive rounds of selection/ampliﬁcation are then performed against the desired target only. No further No only. immobilized on magnetic beads, with the enriched mRNA libraries translated in the same FIT system at each step. In their simplest incarnation, such selections can be performed using a FIT system that primarily incorporates canonical amino acids, initiated by a non- vayinstitutes.be/ canonical d-orl-amino acid bearing a chloroacetyl moiety at the N-terminus and PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

Fig. 2. RaPID system (mRNA display coupled with FIT system). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 241

Platform Technologies for the Artiﬁcial Pseudo-Natural Product Discovery 241

a downstream cysteine in order to form a macrocycle through a thioether linkage (as described above). Using such an approach, potent (IC50 ∼100 nM) and isoform- selective cyclic peptide inhibitors of the human AKT2 serine/threonine kinase have

been reported [16]. A similar approach was used to identify macrocyclic peptide

inhibitors (again, active at nM concentrations) of the bacterial drug-transporter pfMATE, a finding of particular significance since proteins of this family are responsible for drug resistance in both bacterial pathogens and malignant cells [17]. It is worth noting that this particular work involving 3-dimensional X-ray structures of the macrocyclic peptide-pfMATE complexes provided the first example of how the series of macrocyclic peptides that are obtained by RaPID selection interact distribution is allowed. distribution with a target protein.

Selections using libraries including more diverse non-canonical amino acid residues have also been reported. For example, cyclic peptide inhibitors of the 9180 human Sirtuin2 lysine deacetylase were identiﬁed from a library of macrocyclic only. No further No only. peptides engineered to include a triﬂuoroacetyl-lysine “warhead” that mimics the acetyl-lysine substrate of the enzyme [18]. The derived Sirtuin2 inhibitors exhibited high potency (KD as well as IC50 in the single digit nM range) and a high degree of isoform selectivity.

vayinstitutes.be/ RaPID selection using still more complex non-canonical libraries is also possible. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Taking advantage of the capacity of the FIT system to translate -methylated amino acids, a library engineered to express N-methylated Phe, Ser, Gly and Ala, as well as d-Trp (and macrocyclized through an intramolecular reaction between http://www.sol an N-terminal chloroacetyl group and a downstream cysteine as described above) was screened for inhibitors of the human oncoprotein E6AP [19]. The macrocyclic host on on host E6AP inhibitors obtained each included multiple N-methylated residues, and were arguably more like naturally occurring non-standard macrocyclic peptides than any other non-natural macrocyclic peptide inhibitor discovered to date. Moreover, they exhibited extremely high aﬃnity for E6AP, with KDs in the sub-nM to single digit

http://www.worldscientific.com/worldscibooks/10.1142/ nM range, and appeared able to disrupt the protein-protein interaction between E6AP and the P53 tumor suppressor protein.

Outlook to future developments of research on biologicals and biosimilars The RaPID system is quite robust for selecting macrocyclic peptides against var-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World ious therapeutic targets. However two major issues remain to be solved. First, despite the fact that selected macrocyclic peptides are generally highly active and NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY selective against the target proteins, they are not necessarily membrane permeable and oral available even though N-methylation of backbone is implemented into the

selection scheme. It seems critical to improve the RaPID methodology that involves membrane permeability selection step. Moreover, peptidic natural products contain more diverse kinds of non-peptidic structures, such as azolines and azoles, modiﬁed by post-translational modifying enzymes. These non-peptidic modiﬁcations into the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 242

242 H. Suga

peptide backbone might signiﬁcantly contribute to the drug-like properties. There- fore, it is also important for the future methodology development to include such post-translational modiﬁcation using appropriate enzymes.

Acknowledgments This work was supported by a JSPS Grants-in-Aid for the Specially Promoted Research (21000005), JST CREST, and MEXT Platform for Drug Discovery, Infor- matics, and Structural Life Science to H.S. distribution is allowed. distribution References

1. A. H. Barnett, D. R. Owens, Lancet 349, 47 (1997).

9180 2. R. L. Koretz, M. Pleguezuelo, V. Arvaniti, P. Barrera Baena, R. Ciria et al., Cochrane Datab. Syst. Rev. 1, CD003617 (2013). only. No further No only. 3. A. Takeda, K. Cooper, A. Bird, L. Baxter, G. K. Frampton et al., Health Technol. Assess. 14, 1 (2010). 4. R. E. Handschumacher, M. W. Harding, J. Rice, R. J. Drugge, D. W. Speicher, Science 226, 544 (1984). vayinstitutes.be/ 5. D. S. Perlin, Future Microbiol. 6, 441 (2011). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 6. E. M. Driggers, S. P. Hale, J. Lee, N. K. Terrett, Nat. Rev. Drug Discov. 7, 608 (2008).

http://www.sol 7. Y. Goto, T. Katoh, H. Suga, Nat. Protoc. 6, 779 (2011). 8. H. Murakami, A. Ohta, H. Ashigai, H. Suga, Nat. Meth. 3, 357 (2006).

host on on host 9. T. Kawakami, H. Murakami, H. Suga, J. Amer. Chem. Soc. 130, 16861 (2008). 10. T. Kawakami, H. Murakami, H. Suga, Chem. Biol. 15, 32 (2008). 11. T. Fujino, Y. Goto, H. Suga, H. Murakami, J. Amer. Chem. Soc. 135, 1830 (2013). 12. Y. Goto, A. Ohta, Y. Sako, Y. Yamagishi, H. Murakami, H. Suga, ACS Chem. http://www.worldscientific.com/worldscibooks/10.1142/ Biol. 3, 120 (2008). 13. Y. Goto, H. Suga, J. Amer. Chem. Soc. 131, 5040 (2009). 14. A. Ohta, H. Murakami, E. Higashimura, H. Suga, Chem. Biol. 14, 1315 (2007). 15. T. Passioura, H. Suga, Chemistry 19, 6530 (2013). 16. Y. Hayashi, J. Morimoto, H. Suga, ACS Chem. Biol. 7, 607 (2012). 17.Y.Tanaka,C.J.Hipolito,A.D.Maturana,K.Ito,K.Kurodaet al., Nature 496, 247 (2013). ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 18. J. Morimoto, Y. Hayashi, H. Suga, Angew. Chem. Int. Ed. Engl. 51, 3423 NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY (2012). 19. Y. Yamagishi, I. Shoji, S. Miyagawa, T. Kawakami, T. Katoh et al., Chem.

Biol. 18, 1562 (2011). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 243

243

R R

ANTICALINS & PASYLATION : NEW CONCEPTS FOR BIOPHARMACEUTICAL DRUG DEVELOPMENT FROM PROTEIN DESIGN

ARNE SKERRA Lehrstuhl für Biologische Chemie, Technische Universität München

distribution is allowed. distribution Emil-Erlenmeyer-Forum 5, 85350 Freising-Weihenstephan, Germany

9180 My view of the present state of research on biologicals (and biosimilars) only. No further No only.

During the past century, since their discovery by Emil von Behring and Paul Ehrlich [1], antibodies (immunoglobulins) have been considered as ideal tools for molecular recognition in the life sciences and biomedical research. The invention of methods

vayinstitutes.be/ for the production of their functional fragments in E. coli [2, 3] has boosted the gen- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

eration of recombinant antibodies with modiﬁed amino acid sequences and altered - properties. Since then, human (or humanized) monoclonal antibodies have become a partic-

http://www.sol ularly successful class of biologics during the past decade [4], complementing conventional biopharmaceuticals that, altogether, currently represent approximately

host on on host one quarter of newly approved drugs [5]. High target speciﬁcity, therapeutic eﬃ- cacy, safety and, consequently, low failure rate during preclinical and clinical development contribute to the medical and commercial success of antibodies. Thus, immunotherapy has become an established strategy especially for the treatment of

http://www.worldscientific.com/worldscibooks/10.1142/ cancer. However, with regard to novel applications in medicine and biotechnology it appears that immunoglobulins are not always optimally suited, in particular due to their composition of two diﬀerent polypeptide chains and to their very large molecular size. Members of the lipocalin protein family [6] have shown promise as an alternative scaﬀold for the generation of ligand-binding proteins via combinatorial protein design [7, 8]. Natural lipocalins are compact proteins which serve for the transport

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World or storage of biomolecular compounds such as vitamins, hormones and secondary metabolites in many organisms, including humans. Their tertiary structure com- NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY prises a circularly closed eight-stranded anti-parallel β-sheet. This so-called β-barrel supports four structurally hypervariable loops at its open end, which form the en-

trance to the ligand pocket. In comparison with recombinant antibodies, lipocalins provide several advantages as they are composed of a single polypeptide chain, have a much smaller size, and their set of four loops can be easily manipulated at the genetic level, allowing the design of alternative binding proteins that have been dubbed ‘Anticalins’ [7]. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 244

244 A. Skerra

Apart from full size immunoglobulins, most biologics that are successfully used in human therapy, for example interferon, erythropoietin, hormones, growth factors, and recombinant antibody fragments, are relatively small proteins. Their molecular

sizes are much below the threshold for kidney ﬁltration (about 70 kDa), which leads

to rapid elimination from the bloodstream. Accordingly, these therapeutic proteins often have an extremely short half-life in human plasma of just a few hours [9]. This impairs their therapeutic beneﬁt considerably, and in most cases repeated injections at short intervals and high dosing are required in order to achieve a pharmaceutical eﬀect. A currently accepted strategy to prolong the plasma half-life of biologics is their distribution is allowed. distribution chemical conjugation with the water-soluble and inert polyethylene glycol (PEG),

thereby increasing the hydrodynamic volume of the therapeutic protein or peptide beyond the pore size of renal glomeruli [10]. In recent years several PEGylated 9180 biologics have found entry into the clinic, for example PegasysR or PegIntronR , R only. No further No only. both PEGylated versions of interferon α, or Pegﬁlgrastim , a PEGylated G-CSF. However, PEGylation of therapeutic proteins is accompanied by additional effort and high cost: suitable reactive PEG derivatives with high purity are expensive, coupling with the recombinant protein often has poor yield and reduces its activ-

vayinstitutes.be/ ity. Also, additional downstream puriﬁcation steps are needed, wherein product PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- heterogeneity causes further problems. Finally, PEG is not biodegradable and can accumulate in the kidney or other organs upon continued use. Consequently, there is urgent need for other technologies to enable the design of highly active and safe http://www.sol biologics with prolonged action. ‘PASylation’, which makes use of a conformationally disordered and genetically encodable polymer composed of natural amino acids, host on on host provides a biological alternative [11].

My recent research contributions to biologicals (and biosimilars) http://www.worldscientific.com/worldscibooks/10.1142/ Anticalins are a novel class of small and robust engineered protein reagents developed in our laboratory [7, 8], which are usually derived via combinatorial protein design from human plasma lipocalins and oﬀer potential as an alternative to antibodies [12, 13]. The lipocalin protein fold is dominated by a central β-barrel of eight anti-parallel strands, which is open to the solvent at one end where a set of four structurally hypervariable loops forms the entrance to the ligand pocket, similarly

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World to the six complementarity-determining regions (CDRs) of an antibody. While human(ized) immunoglobulins constitute an established class of biotherapeutics today, NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY they show several disadvantages, especially in areas where the full-size molecular format with its immunological eﬀector functions is less relevant. Anticalins have a

much smaller size (160–180 residues), providing better tissue penetration, and they comprise just a single polypeptide chain, thus oﬀering the eﬃcient preparation of fusion proteins and/or conjugation with payloads or enzymes, even by employing inexpensive microbial expression systems. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 245

New Concepts for Biopharmaceutical Drug Development from Protein Design 245

distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Fig. 1. Panel of the X-ray structures of five Anticalins based on the same human lipocalin scaffold that are directed against diverse medically relevant molecular targets: Aβ, Alzheimer amyloid β peptide; CTLA-4, cytotoxic T lymphocyte-associated antigen 4 (CD152); ED-B, extra-domain B http://www.sol of the extracellular matrix protein fibronectin; Hepcidin, a peptide regulator of iron homeostasis; Y-DTPA, yttrium (III) in complex with the chelator [(R)-2-amino-3-(4-aminophenyl)propyl]-trans-

host on on host (S,S)-cyclohexane-1,2-diamine-pentaacetic acid.

Anticalins with speciﬁcities and tight functional blocking activity toward a series of disease-relevant biomolecules have been conveniently generated by targeted

http://www.worldscientific.com/worldscibooks/10.1142/ random mutagenesis of natural lipocalins in combination with phage display selection [14–16]: e.g., the pro-angiogenic factor VEGF-A, the tumor-associated extra- domain B of fibronectin and the deactivating T-cell co-receptor CTLA-4. A beneficial feature of Anticalins is their deep binding pocket, which — similarly to the functions of natural lipocalins — also enables the high affinity complexation of small molecule ligands of various kinds, including toxic substances such as digitalis [17] or radiometal-chelate complexes [15] and even the Alzheimer amyloid β peptide

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World (Fig. 1), thus oﬀering interesting applications in biomedical research, diagnostics and therapy as well as in nuclear medicine or as antidotes. The VEGF-speciﬁc NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Anticalin ‘Angiocal’ has completed Phase I trials for the treatment of solid tumors while the clinical study of another Anticalin, directed against the peptide Hepcidin,

is currently in preparation [13]. Except for full size antibodies, which undergo endosomal recycling, rapid clearance from blood circulation by renal ﬁltration is a typical drawback of therapeutic proteins and peptides, aﬀecting most of all conventional biologics such as hormones, October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 246

246 A. Skerra

growth factors, cytokines, enzymes and almost all clinically relevant antibody fragments (scFv, Fab or single domain/nanobody) or alternative protein scaﬀolds. At present, chemical conjugation with PEG, which expands the eﬀective molecular

size beyond the threshold of kidney ﬁltration, is an established strategy to extend

the short plasma half-life of biologicals to a clinically useful range. However, coupling with the chemical polymer PEG requires expensive material and downstream procedures and, in addition, PEG can accumulate in organs due to its lack of biodegradability. As an alternative, we have developed conformationally disordered polypeptide chains with expanded hydrodynamic volume (Fig. 2) comprising the small amino distribution is allowed. distribution acid residues Pro, Ala, and/or Ser (PAS). PAS sequences are hydrophilic, uncharged

biological polymers with biophysical properties surprisingly similar to PEG [11]. In contrast, PAS polypeptides oﬀer simple fusion to a biologic on the genetic level, 9180 thus permitting facile production in E. coli (or other expression systems) of fully only. No further No only. active proteins in high yields [18] and obviating in vitro coupling or modiﬁcation steps. Also, PAS sequences are biodegradable, thus avoiding tissue accumulation, while showing high stability in blood plasma and lacking toxicity or immunogenicity vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

Fig. 2. Modeled structure of the PASylated Fab fragment of an antibody. The two polypeptide chains of the Fab fragment are colored red while the antigen-binding site is shown in black. 24 arbitrarily selected random conformations of the PAS polymer at the C-terminus of the light chain are presented as snapshots and superimposed in different shades of grey to illustrate its fluctuating random coil-like, space-filling behavior. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 247

New Concepts for Biopharmaceutical Drug Development from Protein Design 247

in animals. Furthermore, by adjusting the length of the PAS tag, typically between 200 and 1000 residues, the eﬀect on plasma half-life can be easily tuned, allowing extension by one to two orders of magnitude.

Outlook to future developments of research on biologicals (and biosimilars) The growing ﬁeld of modern pharmaceuticals is dominated by chemical (small molecule) drugs on the one hand and biologics on the other, which currently contribute approximately one quarter of newly approved substances, with rising overall

distribution is allowed. distribution proportion [5]. Notably, the vast majority of biologics constitutes therapeutic pro-

teins while other biomacromolecules, like carbohydrates or nucleic acids, are con-

ﬁned to niche applications. Peptides may be assigned to both classes, depending 9180 on whether they are manufactured by chemical synthesis — which is more costly but facilitates introduction of covalent modiﬁcations — or by way of biotechnology. only. No further No only.

An interesting recent option is the incorporation of non-natural amino acids into biosynthetic proteins via expansion of the genetic code [19]. This opens the route to the site-speciﬁc introduction of novel side chain functionalities that can either be useful per se, as in the case of spectroscopic probes [20], or may enable the vayinstitutes.be/ eﬃcient and directed conjugation of a protein with any other chemically function- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- alized compound, for example radionuclide chelators or small molecule toxins. So far, drug development has focused on the speciﬁc recognition and/or tight blocking

http://www.sol of disease-relevant cellular/biomolecular targets, which explains the success of antibody technology up to now. Meanwhile, it has become apparent that the careful

host on on host design of additional pharmaceutical properties, in particular biodistribution, circulation half-life and charging with payloads, will be crucial to provide the next wave of biologics — possibly of a hybrid nature — that offer superior therapeutic effects. Eventually, this development could form a fruitful link between the disciplines of chemistry and biology, which have largely separated since the early days of Ehrlich http://www.worldscientific.com/worldscibooks/10.1142/ and von Behring, to yield novel biomolecular drugs for the benefit of human health.

Acknowledgments The author wishes to thank the Deutsche Forschungsgemeinschaft, the Bundesmin- isterium f¨ur Bildung und Forschung, Germany, the European Commission as well as Pieris AG and XL-protein GmbH, both in Germany, for ﬁnancial support. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY References 1. F. Bosch, L. Rosich, Pharmacology 82, 171 (2008).

2. A. Pl¨uckthun, A. Skerra, Meth. Enzymol. 178, 497 (1989). 3. A. Skerra, A. Pl¨uckthun, Science 240, 1038 (1988). 4. J. M. Reichert, MAbs 3, 76 (2011). 5. G. Walsh, Nat. Biotechnol. 28, 917 (2010). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 248

248 A. Skerra

6. B. Akerstr¨˚ om, N. Borregaard, D. A. Flower, J.-S. Salier, Lipocalins. Landes Bioscience, Georgetown, Texas (2006). 7. A. Skerra, Biochim. Biophys. Acta 1482, 337 (2000). 8. G. Beste, F. S. Schmidt, T. Stibora, A. Skerra, Proc. Natl. Acad. Sci. USA 96,

1898 (1999). 9. J. H. Lin, Curr. Drug Metab. 10, 661 (2009). 10. G. Pasut, F. M. Veronese, J. Control. Release 161, 461 (2012). 11. M. Schlapschy, U. Binder, C. B¨orger, I. Theobald, K. Wachinger et al., Protein Eng. Des. Sel. 26, 489 (2013). 12. A. Skerra, Curr. Opin. Chem. Biol. 7, 683 (2003). distribution is allowed. distribution 13. C. T. Mendler, A. Skerra, Drug Future 38, 169 (2013).

14. M. Gebauer, A. Schiefner, G. Matschiner, A. Skerra, J. Mol. Biol. 425, 780

9180 (2013). 15. H. J. Kim, A. Eichinger, A. Skerra, J. Am. Chem. Soc. 131, 3565 (2009). only. No further No only. 16. D. Sch¨onfeld, G. Matschiner, L. Chatwell, S. Trentmann, H. Gille et al., Proc. Natl. Acad. Sci. USA 106, 8198 (2009). 17. F. Eyer, W. Steimer, T. Nitzsche, N. Jung, H. Neuberger, C. M¨uller, M. Schlap- schy,T.Zilker,A.Skerra,Toxicol. Appl. Pharmacol. 263, 352 (2012).

vayinstitutes.be/ 18. S. Di Cesare, U. Binder, T. Maier, A. Skerra, BioProcess. Int. 11, 30 (2013). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 19. T. S. Young, P. G. Schultz, J. Biol. Chem. 285, 11039 (2010). 20. S. M. Kuhn, M. Rubini, M. A. M¨uller, A. Skerra, J. Amer. Chem. Soc. 133, 3708 (2011). http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 249

249

FROM NATURAL ANTIBODIES TO SYNTHETIC PROTEINS

SACHDEV S. SIDHU Banting and Best Department of Medical Research & Department of Molecular Genetics University of Toronto, 160 College Street, Toronto, Ontario, M5S 3E1, Canada distribution is allowed. distribution My view of the present state of research on biologicals and

biosimilars 9180 The human genome project has been a success in terms of providing basic knowledge only. No further No only.

of cellular function. Advanced DNA sequencing technologies and technologies for the systematic analysis of genomes, transcriptomes and proteomes have provided unprecedented insights into the inner workings of both healthy and diseased cells. Consequently, the research community is poised to produce tremendous advances

vayinstitutes.be/ in therapies for numerous human diseases, if only this basic knowledge can be PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

translated to the practical level of drug development. - However, it remains unclear how the vast data acquired from genomics and systems biology can be applied to the practical business of drug development. In

http://www.sol my opinion, a major cause for this disconnect is the lack of advanced tools for the manipulation of proteins with anything close to the speed and scope that is possible

host on on host for the manipulation of nucleic acids. Proteins are central to essentially all cellular processes, most diseases are caused by protein malfunction, and most drugs act by modulating the activities of proteins. Thus, it seems clear that basic knowledge from genomics research can only be translated into meaningful therapies if we can develop

http://www.worldscientific.com/worldscibooks/10.1142/ comprehensive tools for modulating proteins with the same speed and precision that is commonplace for nucleic acids. Such a toolkit would exponentially expand our knowledge of normal and diseased cell function, and most importantly, the tools could be converted into therapeutics that prevent or reverse disease by correcting aberrant protein function. Unfortunately, despite their undeniable importance, current methods for the generation of protein modulators are not scalable to the genome level. Small

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World molecules remain invaluable as both research tools and drugs, but are limited to a small niche of druggable targets and require extensive eﬀort and expense to produce. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Antibodies oﬀer an attractive alternative but they are restricted to extracellular targets and traditional hybridoma technologies that rely on animal immunization are

not scalable in a facile manner. In my opinion, there is an urgent need for advanced “synthetic” protein technologies that take advantage of in vitro selection technologies and man-made protein libraries. With such designed systems, it would be possible to eﬃciently scale up aﬃnity reagent production to a genome-wide scale and October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 250

250 S. S. Sidhu

the binding proteins could be engineered with tailored characteristics that would make them ideally suited for biotechnology applications.

My recent research contributions to biologicals and biosimilars

My group has focused on combining knowledge of protein structure and function with combinatorial phage display technology. Our goal has been to not only develop combinatorial protein libraries capable of generating aﬃnity reagents against diverse antigens, but also, to do so in a deﬁned and controlled manner so that we can gain further insights into library design and develop further improved libraries. The distribution is allowed. distribution

ultimate aim has been to produce synthetic proteins that are simple and highly

stable, but are comparable or better than natural antibodies in terms of aﬃnities and speciﬁcities. Over the course of the past decade, we have progressed from 9180 synthetic antibody libraries, to libraries of synthetic proteins built on alternative only. No further No only.

scaﬀolds, to small proteins that can be produced by chemical synthesis. Our phage-displayed libraries of synthetic antibodies use entirely man-made antigen-binding sites and thus circumvent the need for natural immune repertoires. We have taken a particularly simple approach by using a single, optimized human

vayinstitutes.be/ framework and diversifying only a limited subset of positions within the antigen- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

binding site that are most important for antigen recognition. Moreover, we have - also limited chemical diversity to those amino acids that are most suitable for mediating molecular recognition, including large aromatic residues to provide contacts

http://www.sol and small residues to provide conformational ﬂexibility [1, 2]. The resulting libraries have proven to be extremely functional and have generated thousands of

host on on host high aﬃnity antibodies against hundreds of antigens [3, 4] (Fig. 1(a)). Importantly, the in vitro selection technology has enabled controlled selections to tackle molecular recognition tasks that would be extremely diﬃcult if not impossible with animal immunization, including development of antibodies against protein conformational

http://www.worldscientific.com/worldscibooks/10.1142/ states [5], membrane proteins [6], RNA [7, 8] and post-translational modiﬁcations [9]. Also, the in vitro nature of the technology has made it possible to scale up the selection process to enable antibody generation against hundreds of antigens in parallel. Thus, synthetic antibody technology has now surpassed both the throughput and quality of hybridoma technology and has enabled numerous applications beyond the scope natural antibodies. As a further extension of in vitro library technology, we have gone beyond anti-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World bodies to develop alternative scaﬀolds adapted to tasks for which antibodies are not suitable. In particular, antibodies are evolved for extracellular function and gener- NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY ally do not fold in cells, thus limiting their use for applications that require protein function in an intracellular environment. To address this issue, we have developed

phage display libraries of small proteins that are adapted for intracellular function and have engineered these for enhanced aﬃnity against targets that the natural proteins bind to only weakly. In one example, we have engineered superbinder SH2 domains that bind to phospho-tyrosine peptides with aﬃnities many orders of October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 251

From Natural Antibodies to Synthetic Proteins 251

(a) (b) (c) (d)

distribution is allowed. distribution

9180 Fig. 1. Synthetic proteins in complex with natural ligands. The structures of synthetic proteins and their natural ligands are shown as grey surfaces or cyan ribbons, respectively. (a) A synthetic

only. No further No only. antigen-binding fragment bound to a tudor domain (PDB entry 3PNW). (b) A superbinder SH2

domain bound to a phospho-tyrosine peptide (PDB entry 4F5B). (c) An ubiquitin variant bound to an ubiquitin speciﬁc protease (PDB entry 3MTN). (d) A D-amino acid protein bound to vascular endothelial growth factor (PDB entry 4GLS). vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

magnitude higher than those of natural SH2 domains [10]. This was achieved by - optimizing the SH2 domain residues that contact the phospho-tyrosine side chain, and thus, these domains exhibit enhanced aﬃnity without alterations in speciﬁcity

http://www.sol (Fig. 1(b)). We have used superbinder SH2 domains as inhibitors of natural SH2 domain interactions and have shown that these inhibitors can halt cancer cell growth

host on on host [10], or alternatively, we have replaced natural SH2 domains with superbinders in the context of whole proteins to alter signaling pathways [11]. In another example, we have used ubiquitin as a scaﬀold to generate variants that bind tightly and speciﬁcally to ubiquitin modifying enzymes [12] (Fig. 1(c)). Using selective ubiq-

http://www.worldscientific.com/worldscibooks/10.1142/ uitin variants, we have been able to develop inhibitors of several deubiquitinases and have shown that these inhibitors are functional in live cells. We extended the strategy to target ubiquitin conjugating enzymes and ubiquitin ligase enzymes and, in some cases, showed that ubiquitin variants can also act as enzyme activators. Finally, we showed that the approach can be used to derive binders to non-catalytic ubiquitin-binding domains that function as mediators of protein-protein interactions. Thus, it has become clear that ubiquitin, SH2 domains and other simple

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World intracellular proteins can be engineered as selective inhibitors to modulate protein function inside cells. These advances will open up many avenues for basic research NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY and could also lead to the development of next generation therapeutics. Taking the concept of synthetic proteins a step further, we have developed small,

optimized scaﬀolds that function like antibodies but are amenable to full chemical synthesis, thus enabling the incorporation of non-natural amino acids. The power of the technology has been demonstrated by the development of D-amino acid proteins that act as inhibitors of natural vascular endothelial growth factor [13] (Fig. 1(d)). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 252

252 S. S. Sidhu

Thus, we showed that D-proteins can recognize natural L-proteins with high aﬃnity and speciﬁcity. However, unlike L-proteins, D-proteins are resistant to proteolysis by natural proteases and should be much more stable in serum and cellular environ-

ments. Moreover, D-peptides are not recognized by the major histocompatibility

complex, and thus, they should exhibit lower immunogenicity. Finally, the use of chemical synthesis also facilitates the incorporation of many other non-natural amino acids with special functions and the incorporation of specialized chemical moieties to enable advanced features such as imaging and drug delivery. In sum, these advances in the design of synthetic binding proteins extend the applications for aﬃnity reagents well beyond the range of natural antibodies and distribution is allowed. distribution this should have a transformative eﬀect on many areas of biological research.

9180 Outlook to future developments of research on biologicals and only. No further No only.

biosimilars In the post-genomics era, it has become clear that protein science must evolve if we hope to translate genomics information to the protein level. To meet these demands, protein science must adapt new, rapid methods that rely on precisely engineered

vayinstitutes.be/ synthetic proteins, rather than old, slow methods based on animal immunization PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- and natural antibodies. I believe that continued refinements in synthetic protein library designs will continue to accelerate the pace of affinity reagent generation, and developing binding tools against whole proteomes will soon become a reality. http://www.sol Synthetic approaches and in vitro technologies are also ideally suited to mesh with emerging technologies in nano-scale devices, microfluidics and arrays. Thus, I pre- host on on host dict that not only will we be able to generate synthetic proteins on a proteome scale, but we will also be able to integrate these proteins into next generation systems for cell-based assays and detection. These advances will fuel breakthroughs in basic research and will enable the development of new therapies for unmet medical http://www.worldscientific.com/worldscibooks/10.1142/ needs.

Acknowledgments I thank the Canadian Institutes of Health Research, the Ontario Institute for Can- cer Research, the National Institutes of Health and Reﬂexion Pharmaceuticals for ﬁnancial support. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

References

1.F.A.Fellouse,B.Li,D.M.Compaan,A.A.Peden,S.G.Hymowitzet al., J. Mol. Biol. 348, 1153 (2005). 2. F. A. Fellouse, C. Wiesmann, S. S. Sidhu, Proc. Natl. Acad. Sci. USA 101, 12467 (2004). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 253

From Natural Antibodies to Synthetic Proteins 253

3. F. A. Fellouse, K. Esaki, S. Birtalan, D. Raptis, V. J. Cancasci et al., J. Mol. Biol. 373, 924 (2007). 4. H. Persson, W. Ye, A. Wernimont, J. Adams, R. Lam et al., J. Mol. Biol. 425,

803 (2013).

5. J. Gao, S. S. Sidhu, J. A. Wells, Proc. Natl. Acad. Sci. USA 106, 3071 (2009). 6. S. Uysal, V. Vasquez, V. Tereshko, K. Esaki, F. A. Fellouse et al., Proc. Natl. Acad. Sci. USA 106, 6644 (2009). 7. Y. Koldobskaya, E. M. Duguid, D. M. Schechner, N. B. Suslov, J. Ye et al., Nat. Struct. Mol. Biol. 18, 100 (2011). 8.J.-D.Ye,V.Tereshko,J.K.Frederiksen,A.Koide,F.A.Fellouseet al., Proc. distribution is allowed. distribution Natl. Acad. Sci. USA 105, 82 (2008).

9.K.Newton,M.L.Matsumoto,I.E.Wertz,D.S.Kirkpatrick,J.R.Lill et al., Cell 134, 668 (2008). 9180 10. T. Kaneko, H. Huang, X. Cao, X. Li, C. Li et al., Sci. Sign. 5, ra68 (2012). only. No further No only. 11.G.M.Findlay,M.J.Smith,F.Lanner,M.S.Hsiung,G.D.Gishet al., Cell 152, 1008 (2013). 12. A. Ernst, G. Avvakumov, J. Tong, Y. Fan, Y. Zhao et al., Science 339, 590 (2013).

vayinstitutes.be/ 13. K. Mandal, M. Uppalapati, D. Ault-Riche, J. Kenney, J. Lowitz et al., Proc. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Natl. Acad. Sci. USA 109, 14779 (2012). http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 254

254

FROM INTACT ANTIBODIES TO ARMED ANTIBODIES

DARIO NERI Department of Chemistry and Applied Biosciences, Swiss Federal Institute of Technology (ETH Z¨urich), Wolfgang-Pauli-Strasse 10, CH-8093 Z¨urich, Switzerland distribution is allowed. distribution My view of the present state of research on biologics and

biosimilars 9180 Monoclonal antibodies represent the largest and fastest growing class of pharma- only. No further No only.

ceutical biotechnology products [1, 2]. The main application areas of therapeutic antibodies include cancer, autoimmunity and chronic inﬂammatory conditions. The clinical use of a number of “intact” antibodies (i.e., immunoglobulins in IgG format) is well established and these products are now facing competition by biosimilar

vayinstitutes.be/ products. However, new opportunities are oﬀered by the development of “armed” PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

antibodies, in which the immunoglobulin moiety mainly serves for the selective in - vivo pharmacodelivery of bioactive payloads (such as cytotoxic drugs, bispeciﬁc antibodies, radionuclides or cytokines) to sites of disease, helping increase activity

http://www.sol and spare normal tissues. In this article, I present a personal perspective on the opportunities and challenges associated with the use of armed antibody products

host on on host for the treatment of cancer and inﬂammation. Conventional IgG-based therapeutics may act by a variety of mechanisms, which include the blockade of a biological mediator of disease (e.g., TNF in inﬂamma- tion or VEGF in angiogenesis-related conditions), by the induction of signaling

http://www.worldscientific.com/worldscibooks/10.1142/ events, by the activation of complement or by the recruitment of suitable leukocytes (e.g., NK cells) at the site of disease (a process also called Antibody-Dependent Cell Cytotoxicity, or ADCC). In Oncology, monoclonal antibody products are not terribly eﬃcacious when used as single agents for the treatment of bulky solid tumors. This limitation may be related to the rather ineﬃcient antibody uptake into the neoplastic mass, which is a consequence of limited extravasation rate and of blockade by perivascular tumor

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World cells (the so-called “antigen barrier”). In mice, the best tumor-targeting antibodies may exhibit a tumor uptake as NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY large as 50–100% injected dose per gram (i.e., 5–10% of the injected dose localize to a 100 mg tumor mass, after the initial distribution phase). The corresponding

%ID/g value on solid tumors in cancer patients may be in the order of 0.1–1%, mainly as a consequence of the diﬀerent body weight (20–25 g against 50–100 kg) of the two organisms. Thus, only a relatively small portion of the antibody molecule reaches the target following intravenous administration, while virtually all antibody October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 255

From Intact Antibodies to Armed Antibodies 255

product, at some stage, will pass through an excretory organ (the hepatobiliary system for intact antibodies, the renal system for small antibody fragments). Yet, at suitable time points (e.g., after 24 hours), the best antibody products may exhibit

tumor:blood and tumor:organ ratios greater than 10:1, thus being suitable for the

preferential pharmacodelivery of payloads at the site of disease [3, 4]. For these reasons, antibodies are increasingly being considered as “vehicles” for the targeting of payloads not only to cancer lesions, but also to other conditions (e.g., arthritis, psoriasis, multiple sclerosis, inﬂammatory bowel disorders).

My recent research contributions to biologics and biosimilars distribution is allowed. distribution Antibodies can be used to deliver bioactive molecules (drugs, cytokines, photosen- sitizers, radionuclides, etc.) to the tumor environment, thus sparing normal tissues.

9180 The targeting of modiﬁed sub-endothelial extracellular matrix components using armed antibodies is particularly attractive, because of: (i) the abundance and sta- only. No further No only. bility of some of these antigens; (ii) the fact that some of these antigens are strongly expressed in various cancer types and in angiogenesis-dependent diseases, while being virtually undetectable in most normal adult tissues; (iii) the accessibility of these structures from the blood-stream. The use of armed antibodies for the targeting of vayinstitutes.be/ newly-formed blood vessels is a therapeutic strategy, often referred to as “vascular PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- targeting” [4]. Over the last 20 years, we have demonstrated the feasibility of antibody-based vascular targeting strategies in several mouse models of cancer [4], arthritis [5], http://www.sol endometriosis [6], angiogenesis-related ocular conditions [7], atherosclerosis [8], vas- culopathy [9], inﬂammatory bowel diseases and psoriasis [unpublished]. Five armed host on on host antibody products developed in my lab have been moved to Phase I and Phase II clinical trials, in collaboration with industrial partners, for the treatment of cancer and of rheumatoid arthritis. We have explored the role of cytotoxic agents [10, 11], radionuclides [12] and http://www.worldscientific.com/worldscibooks/10.1142/ cytokines [13] as possible payloads for the development of armed antibody products. In particular, antibody-cytokine fusion proteins (immunocytokines) exhibit a considerable therapeutic potential and a number of favorable properties: (i) these products can be very active in patients at low (milligram) doses; (ii) cytokines have often already been studied in the clinic, thus facilitating the development of the corresponding immunocytokine products; (iii) immunocytokines can be used to po- tentiate or attenuate the action of the immune system at the site of disease and are ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World easy to use in combination with other drugs; (iv) unlike antibody-drug conjugates, NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY these products do not exhibit a direct damage of clearance-associated organs.

Outlook to future developments of research on biologics and

biosimilars Even though biosimilars are not identical to the original product (they may diﬀer in terms of glycosylation patterns and, potentially, also in terms of microhetero- October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 256

256 D. Neri

geneities and diﬀerent formulation), the recent approval of biosimilar antibody products indicates that this class of medicinal agents is already a pharmaceutical reality, which may receive a broad indication label after the execution of a limited

amount of clinical studies. Pharmaceutical companies may prefer to focus on the

development of innovative antibody products, since break-through technologies and novel therapeutic concepts now allow the generation of products of unprecedented activity and selectivity.

Acknowledgments distribution is allowed. distribution

The research of the group Neri is supported by the ETH Z¨urich, the Swiss National

Science Foundation, Oncosuisse, the European Union (FP Program PRIAT and IMI Project QUIC-CONCEPT), the Swiss Federal Commission for Technology and 9180 Innovation (KTI MedTech Award) and by research contracts with the Philogen

only. No further No only. group.

References 1. G. Walsh, Nat. Biotechnol. 28, 917 (2010). vayinstitutes.be/ 2. P. J. Carter, Nat. Rev. Immunol. 6, 343 (2006). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 3. G. L. Poli, C. Bianchi, G. Virotta, A. Bettini, R. Moretti et al., Cancer Im- munol. Res. 1, 134 (2013).

http://www.sol 4. D. Neri, R. Bicknell, Nat. Rev. Cancer 5, 436 (2005). 5. K. Schwager, M. Kaspar, F. Bootz, R. Marcolongo, E. Paresce et al., Arthritis

host on on host Res. Ther. 11, R142 (2009). 6. K. Schwager, F. Bootz, P. Imesch, M. Kaspar, E. Trachsel et al., Hum. Reprod. 26, 2344 (2011). 7. M. Birchler, F. Viti, L. Zardi, B. Spiess, D. Neri, Nat. Biotechnol. 17, 984 (1999). http://www.worldscientific.com/worldscibooks/10.1142/ 8. M. Fiechter, K. Frey, T. Fugmann, P. A. Kaufmann, D. Neri, Atherosclerosis 214, 325 (2011). 9. M. Franz, I. Hilger, K. Gr¨un, S. Kossatz, P. Richter et al., J. Heart Lung Transpl. 32, 641 (2013). 10. G. J. Bernardes, G. Casi, S. Tr¨ussel, I. Hartmann, K. Schwager et al., Angew. Chemie Int. Ed. Engl. 51, 941 (2012). 11. N. Krall, J. Scheuermann, D. Neri, Angew. Chemie Int. Ed. Engl. 52, 1384 ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World (2013). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 12. M. Steiner, D. Neri, Clin. Cancer Res. 17, 6406 (2011). 13. N. Pasche, D. Neri, Drug Discov. Today 17, 583 (2012).

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 257

257

REGULATING CELLULAR LIFE DEATH AND DEVELOPMENT USING INTRACELLULAR COMBINATORIAL ANTIBODY LIBRARIES

RICHARD LERNER, JIA XIE, HONGKAI ZHANG and KYUNGMOO YEA Department of Molecular Biology, The Scripps Research Institute distribution is allowed. distribution

JOEL BLANCHARD and KRISTIN BALDWIN

Department of Neuroscience, The Scripps Research Institute 9180 Recently we developed a method where large antibody libraries are rendered infec- only. No further No only.

tious for eukaryotic cells, thereby directly linking the antibody genotype to the phenotype of individual cells. The power of the method derives from the fact that the potential agonists are selected from a large na¨ıve or preselected diversity system that contains about 108–1011 unique members. In the best case when autocrine screen- vayinstitutes.be/ ing systems are employed, one can analyze about two million phenotypic events per PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- hour. Using these methods, a variety of agonist antibodies have been discovered that regulate cell fates, including those that are phenocopies of important cytokines

http://www.sol as well as those that regulate the lineage speciﬁcation and transdiﬀerentiation of stem cells. The method only seems to be limited by the availability of selection

host on on host systems. One of the most important cellular fates yet to be explored is that of survival from death which is, arguably, the most selectable of all possible phenotypes. We began our studies on the selection of antibodies that inhibit cell death by a study of cell survival after virus infection (Fig. 1). http://www.worldscientific.com/worldscibooks/10.1142/ One can imagine three selection formats for such studies. In the first, one uses prior knowledge to hypothesize which proteins if perturbed would inhibit the pathogenic effects of viral infection. In this case an intracellular library that was enriched by phage panning or binding to the target protein would be used. In this format one simply asks whether the selection of the target was correct and if antibodies expressed inside cells can disable the target. In the second scenario, one does not presume any prior knowledge of the target. Unbiased libraries are ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World employed and the selection becomes a discovery tool for finding molecules that NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY are critical to pathogenic pathways. Finally, when the unbiased format is used, one can follow the discovery rounds by subsequent selection rounds that enrich the

libraries against the newly identiﬁed protein to validate the discovery and ﬁnd more antibodies that interact with the target. Here, we will illustrate these principles by showing how unbiased combinatorial antibody libraries expressed inside cells can be used to identify molecules that, when inactivated, prevent virus induced cell death. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 258

258 R. Lerner et al.

distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

http://www.sol Fig. 1. Scheme of the phenotype selection based on protection from cell death. The selection starts with a na¨ıve intrabody Lentiviral library that when co-infected with human Rhinovirus in HeLa cell cultures allows selection of antibodies that protect a few cells from death. The integrated host on on host antibody fragments (the information) from the cells that survived were recovered and converted to a secondary Lentiviral library which is the starting point for the second round of selection. The selection was carried out for 5 rounds. The ratio of surviving cells (smiley faces) over killed cells (green faces) increases in each round along with functional antibody sequences (red and black hearted particles). http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Fig. 2. Antibodies were used to replace Sox2 and cMyc. These antibodies allowed generation of pluripotent cells and production of live chimeric mice. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 259

Regulating Cellular Life Death and Development 259

While we focused on virus induced cell death as an exemplar, one should be able to extend this method to other systems where there is a selection for survival such as occurs when tumors evade killing by immune cells or small molecules.

In another iteration of the method, we have used intracellular and membrane

libraries to generate antibodies that induce diﬀerentiation of CD34+ bone marrow cells into red cells, platelets, granulocytes, dendritic cells, brown fat cells, and neural cells from a population of. In the perhaps most dramatic example, antibodies were used to replace transcription factors for the generation of pluripotent cells that were capable of generating live mice (Fig. 2). distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 260

260

NANOBODIES: A UNIVERSE OF VARIABLE DOMAINS AND A TOOLBOX FOR MANY TRADES

LODE WYNS Structural Biology Brussels, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussel, Belgium distribution is allowed. distribution

My view of the present state of research on nanobodies as biologicals and research tools 9180 The vertebrate immune system can recognize and respond to an enormous number only. No further No only. of foreign substances. In doing so and in line with the theme of this conference it expands a universe of proteins, in this case a universe of antibodies. In essentially no time it creates a huge repertoire by V,D,J segment recombination and somatic hypermutation. Over a few days neo-Darwinian real-time evolution by an

vayinstitutes.be/ unprecedented process, harnessed in lymphoid organs creates a vast repertoire [1]. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- This natural iterative process of mutation/selection has inspired powerful biotechnological approaches: antibody libraries are presented in optimized formats on a range of display systems. http://www.sol A format presenting unique features is that of camelid VHHs or Nanobodies (Nbs): the variable domain of heavy chain-only antibodies. These antibodies, devoid host on on host of light chains were discovered quite serendipitously in the frame of studies on Trypanosomiasis (sleeping sickness) in cattle and camelids. Despite the absence of light chains, Nbs allow the high aﬃnity binding of antigens providing equivalent- sized binding surfaces as bona ﬁde VH-VL constructs do. They allow binding to http://www.worldscientific.com/worldscibooks/10.1142/ cryptic epitopes and cavities, present new “non-canonical” CDRs, enlarged CDR1 and CDR3 loops and their germline coding segments present several additional hypermutational hotspots. It is obvious that over the last decade these unique features have found their translation in a wide range of studies, both fundamental and applied [2, 3].

My recent research contributions to nanobodies as biologicals and ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World research tools NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Nanobodies as crystallization chaperones The nanobody format has been successfully exploited to help and crystallize recal-

citrant proteins of very different nature. – In an earliest study [4] a specific VHH allowed structure determination of the intrinsically flexible antidote MazE; a short-lived protein antagonizing a stable October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 261

Nanobodies: A Universe of Variable Domains and a Toolbox for Many Trades 261

toxin mediating programmed cell-death in bacteria. All lattice interactions in the crystal exclusively involve Nb-Nb contacts and only 45% of the polypeptide is ordered.

– Having many independent crystal forms has many beneﬁts for structure-based

inhibitor design. Recently the structure of BACE2, an aspartic protease and drug target for the expansion of β-cell mass in diabetes was solved in a series of conformations [5]. A tool chest of BACE2 crystals was created with Fab’s, VHHs, and Fynomers. As they bound to many diverse epitopes and give a range of diﬀerent crystal forms VHHs appear as the best compromise providing bulk to form crystal contacts and suﬃcient stability. distribution is allowed. distribution – S-layers, regular 2D-protein layers constitute major cell wall components in bac-

teria. They have sparked interest in their use as patterning and display scaﬀolds in nano-biotech. Structural data however are quite limited to truncated and 9180 assembly-negative proteins. The Nbs however, used as chaperones, break the only. No further No only. self-polymerizing propensity into 2D lattices. This has allowed the structure determination of full length SbsB (920 residues) and combined with electron density projections from cryo-EM provided the creation of high-resolution maps of the lattices [6]. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/

Fig. 1. Surface representation of the 3D-crystal packing of SbsB S-layer/Nb complex. The inclu- ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World sion of the Nb (magenta) causes mismatch between crystal and S-layer packing. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Nanobodies in studies of amyloidosis

Nanobodies can protect against the formation of pathogenic aggregates at diﬀerent stages in the structural transition of a protein in the soluble native state into amyloid ﬁbrils. Studies on β2microglobulin, α-synuclein and amyloidogenic human lysozyme mutants illustrate their value as structural probes to study these processes. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 262

262 L. Wyns

– Nbs against β2m were used to trap and characterize intermediates in the process leading to formation of β2m ﬁbrils, the cause of dialysis related amyloidosis (DRA) [7]. Nbs were selected that block ﬁbrillogenesis of the crucial ΔN6β2m

protein (β2m deleted of its 6 N-terminal residues). The crystal complex with

Nbs identiﬁes domain swapping as a plausible mechanism of self-association. In the swapped dimer two extended hinge loops — corresponding to the NHVTLSQ heptapeptide — are unmasked and fold into a new minimal two stranded β sheet. The strands of this sheet are prone to associate and stack perpendicularly to the strands direction to build large intermolecular sheets providing an elongation mechanism by self-templated growth. distribution is allowed. distribution – Human lysozyme (HuL) variants (I56T, D67H) are the cause of familial non-

neuropathic systemic amyloidosis. Different Nbs mainly directed against either the HuL α or β domain block fibril formation at different stages along the pathway 9180 [8, 9]. only. No further No only. – Structural studies reveal the epitope for nanobody cAbHuL6 includes neither the site of mutation nor most of the residues in the region destabilized by the mutation. The Nb achieves its effect by binding and fully restoring the structural cooperativity of the wild type protein. It suppresses a local cooperative transition β vayinstitutes.be/ in which the domain and the C-helix unfold. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- – Unlike earlier Nbs, cAbHuL5 binding the domain, does not restore the global cooperativity and transient unfolding of the mutant proteins. Instead it inhibits a subsequent step in the ﬁbril assembly, involving unfolding and reorganization http://www.sol of the α domain. host on on host Nanobodies in trypanosomiasis The discovery of heavy chains-only antibodies occurred in the frame of work on trypanosomiasis. In the diﬃcult combat against sleeping sickness they provide new

http://www.worldscientific.com/worldscibooks/10.1142/ angles of approach. Trypanosomes evade the immune response by continuously changing the VSG’s (Variant Surface Glycoproteins) that cover their entire membrane, which leaves little prospect for a conventional vaccine. Moreover the bloodstream forms are able to turn-over the total surface-exposed VSG pool within 12 minutes through exocytosis in the ﬂagellar pocket. As such this VSG recycling allows antibody shedding and protection.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Nbs raised against VSG’s from diﬀerent Trypanosome species were shown to exhibit peculiar and useful activities toward the parasites. Unlike other antibody NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY formats, VHH’s penetrate the dense VSG coat and are able to target conserved cryptic epitopes [10]. Normal human serum contains apolipoprotein L-I (apoL-I)

which lyses African trypanosomes, except resistant forms. As such T.b.rhodesiense expresses SRA protein endowing this parasite to infect and cause disease in humans. A truncated form Tr-apoL-I was engineered, deleted of its SRA-binding domain. A construct conjugating Tr-apoL-I with a VHH eﬃciently targets conserved cryptic October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 263

Nanobodies: A Universe of Variable Domains and a Toolbox for Many Trades 263

distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host

Fig. 2. Possible mechanism for lysozyme ﬁbril formation and its inhibition by nanobodies cAb- HuL6, cAbHuL22, cAbHuL5.

http://www.worldscientific.com/worldscibooks/10.1142/ epitopes, common to all trypanosome classes, to generate an immunotoxin with potential for trypanosomiasis therapy [11]. In contrast to larger antibody formats a range of VHH’s with nanomolar affinities, both against conserved and class-specific VSG’s, show direct trypanotoxic activities and lyse efficiently parasites independent of complement action. By interfering with the early onset endocytosis they circumvent the classic antibody shedding process. Similar morphological features of the process have been reported to occur upon RNAi silencing of genes involved

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World in clathrin-mediated endocytosis. The exact molecular mode-of-action remains to be elucidated [12]. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Outlook to future developments of nanobodies as biologicals and

research tools The antibody-based and in particular Nb-based libraries, by excellence products of Nature’s use of the “rational design by random design” principle, can for a October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 264

264 L. Wyns

long time remain formidable tools. Development of Nbs as research tools, Nbs for medical imaging, as diagnostics and therapeutics ﬁgures on many research agendas. Innovating, ground-breaking applications are developing at high speed.

– Von Zastrow’s team put them at work as “conformational biosensors that reveal GPCR signaling from endosomes”. Active-state specific Nbs pinpoint where and when GPCR’s signal in cells; video’s provide evidence for different G protein and β-arrestin mediated waves [13]. The potential of Nbs to act as conformational biosensors and to modulate protein-protein interactions opens a world of applications. – In the trypanosome field delivery systems for functional Nbs to the tsetsefly, based distribution is allowed. distribution

on its bacterial endosymbiont Sodalis glossinidius have been developed, oﬀering

perspectives for interfering with the parasite’s life-cycle in the insect [14].

9180 – A system for prophylaxis and therapy against rotavirus disease using transgenic rice expressing rotavirus-speciﬁc VHH’s was developed recently [15]. only. No further No only.

Acknowledgments This work, for a long period, has been supported by grants from VUB, FWO, IWT, vayinstitutes.be/ VIB and EU. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

References http://www.sol 1. S. D. Wagner, M. S. Neuberger, Annu. Rev. Biochem. 76, 1 (2007).

host on on host 2. S. Muyldermans, C. Cambillau, L. Wyns, Trends Biochem. Sci. 26, 230 (2001). 3. E. De Genst, K. Silence, K. Decanniere, R. Loris, J. Kinne et al., Proc. Natl. Acad. Sci. 103, 4586 (2006). 4. R. Loris, I. Marianovsky, J. Lah, T. Laeremans, G. Glaser et al., J. Biol. Chem.

http://www.worldscientific.com/worldscibooks/10.1142/ 278, 28252 (2003). 5. D. W. Banner, B. Gsell, J. Benz, J. Bertschinger, D. Burger et al., Acta Cryst. D 69, 1124 (2013). 6. E. Baranova, R. Franzes, A. Garcia Pino, N. Van Gerven, D. Papastolou et al., Nature 487, 119 (2012). 7. K. Domanska, S. Vanderhaegen, V. Srinivasan, E. Pardon, S. Giorgatti et al., Proc. Natl. Acad. Sci. 108, 1314 (2012).

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 8.M.Dumoulin,A.Last,A.Desmyter,D.Canet,D.Archeret al., Nature 424, 783 (2003). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 9. E. De Genst, P. H. Chan, E. Pardon, S. T. Hsu, J. R. Kumita et al., J. Phys. Chem. B 117, 13245 (2013).

10.B.Stijlemans,K.Conrath,V.Cortez-Retamozo,H.VanXong,L.Wynset al., J. Biol. Chem. 279, 1256 (2004). 11. T. Baral, S. Magez, B. Stijlemans, K. Conrath, B. Van Hollebeke et al., Nat. Med. 12, 581 (2006). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 265

Nanobodies: A Universe of Variable Domains and a Toolbox for Many Trades 265

12. G. Caljon, B. Stijlemans, D. Saerens, S. Muyldermans, S. Magez et al., PLOS Negl. Trop. Dis. 6, e1902 (2012). 13. R. Irannejad, J. C. Tomshine, J. R. Tomshine, M. Chevalier, J. P. Mahoney et

al., Nature 495, 534 (2013).

14. L. De Vooght, G. Caljon, B. Stijlemans, P. De Baetselier, M. Coosemans et al., Microb. Cell Fact. 11, 23 (2012). 15. D. Tokuhara, B. Alvarez,´ M. Mejima, T. Hiroiwa, Y. Takahashi et al., J. Clin. Invest. 123, 3829 (2013). distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 266

266

SESSION 5: BIOLOGICALS AND BIOSIMILARS

CHAIR: MARKUS G. GRUTTER¨ AUDITORS:A.LESAGE1, S. MUYLDERMANS2 (1) GrayMatters Consulting, Nachtegalendreef 27, 2980 Halle-Zoersel, Belgium (2) Structural Biology Research Center, VIB, and Research Group of Cellular and Molecular Immunology, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium distribution is allowed. distribution

9180 Discussion among panel members

only. No further No only. Markus Gr¨utter: May I start with a question to Richard Lerner: I was astonished

by hearing, although not being in this area, that you can take a particular cell and reprogram it with a single antibody into a fat cell or diﬀerentiate it into any other cell. I always thought this involved a whole machinery of reprogramming. How does this go with just one molecule in the cell? vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Richard Lerner: Actually in the normal situation, it doesn’t require a whole ensemble. Because if you look at the normal hormones like erythropoietin or GCSF

http://www.sol or whatever, a single molecule can take a stem cell and cause it to run a particular route. But what is surprising is, which you say is true, a single antibody can cause

host on on host a stem cell to run a route it is not supposed to run. And I think that has to do with what we were discussing either yesterday or the day before that these receptors can be used in more than one way, because of the dwell time. An antibody is not what they are used to seeing.

http://www.worldscientific.com/worldscibooks/10.1142/ Sachdev Sidhu: I continue on a mechanistic level. Do you see most of these dramatic eﬀects come from agonists? Maybe we have been too obsessed with developing antagonists with antibodies.

Richard Lerner: All of the 20 (I did not show them all) are agonists. However there is a paper which is shortly out by McCaﬀerty in Cambridge, and he actually makes an antagonist. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Sachdev Sidhu: That is interesting. Because in therapeutic development we are NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY trying to basically throw away most agonists antibodies and focus on the inhibitors I think.

Richard Lerner: If you had not thrown away all your agonists, Genentech would have been successful! October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 267

Session 5: Discussions 267

Sachdev Sidhu: They are turning over as we speak.

Sachdev Sidhu: I have a question for Dr. Suga. Very cool approach. Have you

applied it at all to larger proteins, I guess it could be suitable for that?

Hiroaki Suga: Yes, my lab has probably done 20 targets with maybe 5 or 6 of them being membrane proteins. The company that I have actually has done more than 30 diﬀerent targets. So you can basically target anything. And for membrane proteins, we often have to screen against which one has a biological activity. Enzymes in general are very easy to get inhibitors for very quickly. distribution is allowed. distribution Richard Lerner: Professor Wyns, can I ask a question? One of the interesting things I think is, that in infectious diseases, like Trypanosomes, the evolution of the surface

9180 of the parasite has partly considered “evasion” of the immune system. It is used to evading an ordinary antibody but you showed that with this unusual antibody only. No further No only. construct it cannot evade it. But in an organism like the llama, where you got these heavy chain antibodies (nanobodies), its own viruses, its own parasites, would have to have learned to evade that highly specialized system, right? It does not have to evade a human antibody, it has to evade a llama nanobody.

vayinstitutes.be/ So my guess is if you isolate the viruses or parasites from that organism, they PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- would be able to evade the heavy chain only antibodies. Because they would have learned to evolve its parasites to escape the immune system, just like inﬂuenza virus has learned how to do it in man. There must be all kind of infectious agents for http://www.sol that beast.

host on on host Lode Wyns: I just skipped one of the slides showing that there are opportunities of multiple types of attack, with the antibodies, as I said: antibodies that sneak in through the VSGs (Variant Surface Glycoproteins) of the Trypanosome.Youcan create antibodies of a whole range now that can be taken up. . . http://www.worldscientific.com/worldscibooks/10.1142/ Richard Lerner: But that is not the issue. The issue is that you evolve a parasite in part to evade the immune system. The way parasites will evade that scaﬀold will be diﬀerent than the way you evade an ordinary antibody. You could test that, if youwantedto.

Lode Wyns: You have always this race between what you do and a what a parasite ﬁnds as a solution but. . . ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Richard Lerner: But the parasite is up against the immune system in the organism where it lives. It has got to specialize to evade that particular kind of antibody.

Lode Wyns: But, there is the other question around I think: why did the camels evolve these VHH? And I don’t know this. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 268

268 Biologicals and Biosimilars

Richard Lerner: It is obviously a function of the hump!

Markus Gr¨utter: I have a question to Dario Neri. You showed nicely that you load

antibodies but also single chains Fv’s, shortened pieces with all kinds of interesting payloads and then get it to the target and cure diseases. What is the lifetime of

these smaller pieces, these single chain Fv’s, in the body?

Dario Neri: We typically work with either bivalent or homotrivalent derivatives, and they all clear very rapidly. Like the half-life in the mouse is about 20-30 minutes, in human it may be 1-2 hours. This is what you want if you have a payload like a pro-inﬂammatory cytokine that should not stay in blood for a long period of distribution is allowed. distribution

time. So in that particular situation, you want a long residence time at the site of

disease — ideally weeks — and you want a rapid clearance from the blood. I think

9180 the approach has a potential for certain indications but a more general problem is that antibodies and proteins in general are very slow at getting out of blood only. No further No only. vessels. That is even though the pictures look nice at late time points and you have a selectivity, it is still a minute portion of what you inject that reaches the target. This is really a problem because most of the diseases happen outside of blood vessels and you have to get there. So I think part of the fascination is trying to get smaller vayinstitutes.be/ molecules like Hiroaki Suga showed or by using DNA-encoded chemical libraries. In PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- our experience, when molecules get smaller than 2000 Da, extravasation can be very rapid, it can happen in a matter of seconds, then all of a sudden we are switching the

http://www.sol targeting game from hours and days back to minutes. Of course, this will not work for all payloads but it may work for example for cytotoxics with suitable cleavage

host on on host sites.

Richard Lerner: Can I raise a scale problem because this comes up every time? This could be quite unpopular. Most of the people who worry about making an antibody small have never seen a blood vessel. I don’t think there is any capillary http://www.worldscientific.com/worldscibooks/10.1142/ that is so small that a full antibody cannot get through it. It is sort of like saying “Iamgoingtobuildabridge60feethighsoa7footmasscangetthroughit”.If you look at the scale of the vessel tube relative to the molecular dimensions of the antibody molecule, the ratio is very large. So I would like to challenge the concept of “Small is beautiful” in the antibody world.

Dario Neri: My experience and the experience of many groups is that antibodies

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World eventually get to the perivascular targets, but it is a very slow process. You can

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY measure that very precisely by quantitative biodistribution studies. Unfortunately, actually most companies do not do enough imaging for my taste but quantitative biodistribution studies show very clearly that if you wait 2-3-4 days you get a target to non-target ratio of 10 to 1 (20 to 1), but the process is very slow. So the question is really, does it matter? It depends on the payload. If you have an intact antibody and your antibody is doing something, that is not a problem. It will swim happily October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 269

Session 5: Discussions 269

in the blood until it gets to the target and does its job. But for certain applications which are very popular today, like antibody-drug conjugates, it may matter whether the bomb is swimming in our body for 2-3-4 days without doing its job. So I think

that tailoring of pharmacokinetic properties, and Arne Skerra also showed it nicely,

really depends on the therapeutic strategy. Probably diﬀerent strategies will have diﬀerent applications.

Richard Lerner: I agree with that 100%. But that is not a direct function of the size of the molecule. Sure if you cut up an antibody into small pieces, it behaves very well. But the idea is that “small” will swim better through the vessels! distribution is allowed. distribution

Dario Neri: I swear it is true. And you can do a very simple experiment by doing

intravital ﬂuorescence microscopy: You inject an antibody labeled with ﬂuorescein

9180 and you will see the “swimming” into the vascular structures for hours. You can inject ﬂuorescein as you would do for example for ﬂuorescein angiography. And you only. No further No only. will see the molecule going out of the blood vessels in a matter of seconds. I swear it is true.

Arne Skerra: The point is not so much the penetration into small capillaries but

vayinstitutes.be/ more the penetration of the interstitial space, the narrow liquid zones between PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- densely packed cells. Besides, there are some other barriers that are clearly size- dependent with regard to penetration, for example the lung barrier. There are developments going on to administer biologics by inhalation and there it is abso- http://www.sol lutely clear from published data that the smaller a protein gets the better it crosses the alveolar barrier into the blood. Probably something like that is also true for host on on host the blood-brain barrier (BBB). Mark Dennis from Genentech has published that antibodies penetrate the BBB to an extent of may be 1 to 3%, but this fraction is much larger for smaller molecules. One example is leptin which can go through the BBB, and we also have indication that anticalins can cross the BBB, at least http://www.worldscientific.com/worldscibooks/10.1142/ as far as we can judge from the therapeutic eﬀects of our anti-amyloid β-anticalins in mice.

Discussion among all attendees Markus Gr¨utter: We have time for general discussion. I now invite the panel and the audience to start with questions. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Arne Skerra: I have two technical questions to the amazing talk by Dr. Lerner. One is: Do these antibodies act as intrabodies or as secreted antibodies? And the second question is: What may be the inﬂuence of lentiviral integration to the genome?

Richard Lerner: Thanks for asking. I prepared this question (showing slides). The answer is that for various phenotypes you have diﬀerent expressions. You have some that are intrabodies, some that are targeted to the ER or to the plasma October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 270

270 Biologicals and Biosimilars

membrane. You actually have a library with a diversity of 108 integrated into the plasma membrane. (A. Skerra: How exactly is this going?) I will show you how when it comes up. In terms of the important question about the lentivirus, we are used to a control,

using a ﬂuorescent protein-labeled lentivirus (either red or green), and we have never seen a phenotype, ever, that does not mean we never will, but we have never seen a phenotype solely based on lentiviral integration. And every experiment has a parallel lentiviral control. This is the membrane-bound approach. We call it “nearest neighbor”, and what you see (on the slide) is the G-CSF receptor and the antibody (Ab) integrated in the distribution is allowed. distribution plasma membrane. We call it “nearest neighbor” because it is a ﬁrst approximation

the antibody can only activate what it touches (showing points on the slide). If you

9180 look at the ﬂuorescence of the receptor and the antibody, they wind up in the same place. Basically you get an antigen-antibody union in the plasma membrane. It is only. No further No only. a powerful way to go because the eﬀect of molarity is very high.

Don Hilvert: Could you clarify what the form of the Ab is? Is it the Fv, the Fab, the entire IgG?

vayinstitutes.be/ Richard Lerner: It is a scFv dimer. And if I could add: sometimes that doesn’t PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- matter. One thing you have to realize is that when you do a lentiviral infection with an MOI (multiplicity of infection) of 3, very often you get more than one lentiviral

http://www.sol integration into the genome. So since the Fc is degenerate, you get bispeciﬁc antibodies whether you want them or not. And it turns out that sometimes it doesn’t

host on on host matter and sometimes it is required. For example — this one is published — in the case of the EPO agonist antibodies, only the bispecifics work because you need, shown by Lodisch and others, a symmetry to activate that particular signal transduction domain. And you can’t get the asymmetry with a symmetrical bispecific. So the bispecific thing is sort of interesting. http://www.worldscientific.com/worldscibooks/10.1142/

Kurt Wüthrich: I would like to lead the discussion towards analytical problems with defining biologicals and defining biosimilars relative to biologicals. One of the participants said yesterday that when looking at post-translational modifications, biologicals are a mess. And biosimilars should be a similar mess to be accepted. We may have analytical tools to show that we have a mess but I think we do not have analytical tools to prove that one mess is similar to another mess. Are there ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World any comments on this point? NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Arne Skerra: I am not so sure about that because, for example, the company Roche has launched an entire mass spectrometry department in order to essentially fully characterize the different chemical modifications and glycoforms that occur in monoclonal antibodies. Of course, these are complex molecules so you can only characterize the mixtures and their composition but this to a very detailed extent. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 271

Session 5: Discussions 271

Personally I am convinced that, in the future, if you go to simpler formats, as I explained in my talk, then also the analytics will become simpler because in the end, if we avoid glycosylation or other post-translational modiﬁcations, if we also

avoid the polydispersity of PEG, then we have mass spectra, for example, that only

give one peak and the analytics becomes very simple. Then the only problem that remains is that there are host-cell related impuri- ties. This is a diﬀerent topic, but I recently had a discussion with somebody from the FDA about it and I foresee a situation in maybe 5 or 10 years where the current requirements for biologicals characterization may become more similar to the requirements for chemical drugs characterization. Because fundamentally there is distribution is allowed. distribution not much of a diﬀerence with regards to analytical characterization.

Kurt Wüthrich: One should first find out if the mess is useful, or whether a clean, 9180 uniformly modified product could do the job. only. No further No only. Richard Lerner: You and I already had many discussions about this. This is mostly a political problem, not a scientific problem. Because modern analytical chemistry, as you know better than anyone, has a resolution that is almost at the level of the proton. But there are several competing forces. You have small biotechs saying vayinstitutes.be/ “Our product is a mess and if you don’t copy the mess forget it”. You have Big PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Pharma saying the opposite. And then you have got the strong political pressure to make generics because the cost of these biologicals is so high. The fact on the

http://www.sol matter is that if you want to make up perhaps a fantasy, a scary scenario, about why you must have a mess, you can easily make up that scenario. And there are

host on on host whole departments in charge with making up a reason why if you clean up the mess, it is dangerous. Because it is “big money”.

Arne Skerra: I have a slightly different view. I think there is good reason that in most cases the mess is actually to be avoided because there is one single species http://www.worldscientific.com/worldscibooks/10.1142/ which shows the desired activity. For many antibodies, this is for example the unfucosylated species, which may not be the majority in the older product, but nowadays technologies are evolving to make this more pure, even if there will never be 100% purity. I have actually a different perception also from the regulatory authorities. It is not so much the biotechs that are in favor of the product mixtures, it is the Big Pharma because it makes it then more difficult for newcomers in the field to reproduce the spectrum of a composition. In the end, it is only a consideration of ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World the risk for the patient, so the batch-to-batch reproducibility that has to be secured

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY by applying proper analytical means and by ensuring that the composition is always the same within a certain range. But if we come up with designed biological drugs that are single species drugs, then this problem all of a sudden will disappear.

Richard Lerner: You are talking about going forward. But what I am talking about is if you want to make a new “HUMIRA” (Human Monoclonal Antibody in October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 272

272 Biologicals and Biosimilars

Rheumatoid Arthritis), what are the criteria that the FDA is going to accept. You are absolutely right in going forward. The scientiﬁc community should ﬁgure out a way to make the most homogeneous product possible. That is not the situation we

are in right now. We have multi-billion dollars antibodies that people want to copy

and they will not produce this homogeneous species.

Dario Neri: If I may expand on this topic, the situation at present became very debated because biosimilars were approved for Remicade. Remicade is an antibody (anti-TNF-α) that sells several billions US dollars every year and it is approved for rheumatoid arthritis, psoriatic arthritis, ankylosing spondylitis, psoriasis, IBD distribution is allowed. distribution

(inﬂammatory bowel disease). Now the situation is the following: each approval

for each indication requires the dedicated clinical trials and proof of eﬃcacy. What is not commonly known is that not all TNF blockers, like Remicade, work in all 9180 indications. They typically all work in rheumatoid arthritis and they were ﬁrst

only. No further No only. approved in arthritis, but not all TNF blockers work for example in IBD. Enbrel

(Etanercept) does not, to name one. With the current regulations, the biosimilar is tested in only one indication, typically the easiest, and not at a dose at which the original product was developed, typically a higher dose. But then the product gets automatically approval for all indications, including the diﬃcult ones, for example vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS IBD. So I think this is what will be learned in the near future: either the process

- is too relaxed and maybe the product does not deserve the approval or maybe the small diﬀerences in glycosylation do not have such a big impact. http://www.sol Sachdev Sidhu: I think it is important to make the distinction, as I think Dr. Lerner

host on on host said, “going forward” versus the past. I’d say going forward, I think there is a strong evidence that making the best homogeneous product is best because you are going to have better efficacy and hopefully fewer side effects etc. I think a lot of this does affect that to some degree. Looking back though, I think there are two types of biologics, there is the natural EPO, growth hormone, interferons, and there the http://www.worldscientific.com/worldscibooks/10.1142/ mess is natural, you try to replicate a natural protein. I still would argue the middle ground, which are existing antibodies, are still human made, they are not entities, and I think there is substantial room there for the “bio better approach” — it is kind of a cheesy phrase! If you can find, an example of Retuxan, the subpopulation that is really effective, you then have a chance of going in, as HUMIRA proved versus Enbrel, then subtle advances can really be picked up and I think there is still a lot of play in the existing antibodies, which are historical entities, they are not ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World EPO, they are not evolved proteins. If you can find the subpopulation that is really NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY effective and show that that it is better even a reduction in cost of goods or some side effects, I think that people would switch to the better medicine. I think even

the middle part of the historical stuﬀ is still addressable with more homogeneity, as long as it is a better drug, as a result of that homogeneity.

Lode Wyns: Dr. Lerner was making the problem of the big pharma companies but I really think, through things that I have read in the literature now, that companies October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 273

Session 5: Discussions 273

are playing the double game. On one hand, they may try to block newcomers but at the same time they themselves want to go for biosimilars. They have a problem there as well. Second I think, if you go through the literature, it is always a mess

that is becoming worse, where the mess with new antibodies and new techniques

we have, might become a lesser mess. But I have the impression it all concentrates on the immunogenicity of carbohydrates. We should have a lot of people dealing with this immunogenicity of carbohydrates, that is the core of the matter.

Markus Gr¨utter: I would like to switch the topic from this industry-driven discussion to science applications. You have seen all these scaﬀolds, these new tools that distribution is allowed. distribution

involve a lot of technology libraries with large numbers of molecules from which

speciﬁc binders to targets can be selected. You have seen technologies that are very useful e.g., in the membrane protein structure biology ﬁeld or for target validation. 9180 My question would be: What is the philosophy to make these technologies available

only. No further No only. to the community? From C-H Wong, we have heard that one can approach him

and cooperate at a certain cost. What is the general view on that?

Brian Roth: I can address that but ﬁrst I want to get back to the topic you were trying to steer us away from. Just a comment about the point you made, which vayinstitutes.be/ may or may not be true, that a biological which is highly puriﬁed will have the same PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- eﬃcacy as one which has multiple components. Of course, we know that if it is a protein, then fragments may have subtle or profound diﬀerences at known as well

http://www.sol as at unknown targets, which then could aﬀect the ultimate spectrum of eﬃcacy in a clinical population. That is an experiment or hypothesis which has yet to be fully

host on on host tested. And certainly with antibodies, that is obviously true, because depending on their conformation, they can have profoundly different effects on the ultimate target. But certainly you will have something, which is more highly characterized, but it may actually have a different spectrum of efficacy. I wouldn’t be so strong about that particular point. http://www.worldscientific.com/worldscibooks/10.1142/ In terms of applications, the thing that occurred to me — this sort of makes this answer longer — one of the hats that I wore for a number of years as an adviser for the molecular libraries screening center program (mli.nih.gov/ )which is a huge NIH initiative to discover new small molecules probes, which I think was largely unsuccessful for many reasons. But one of the things they were targeting was protein-protein interactions. These are not so easily targeted by ordinary small molecules but it occurs to me that a lot of the technologies that were discussed are ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World like perfect for that, sort of one of the final frontiers. So I said this as a comment NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY but my question was: some of you have shown a remarkable selectivity within a target class, for example caspases, I guess with the darpins. But have you ever

looked outside this target class? Sort of a broader view of the proteome to see what other targets they might be interacting with through chance interactions?

Richard Lerner: That is the whole reason for using unbiased libraries. By deﬁnition, if you have a target in mind, the library can’t be unbiased because you are panning October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 274

274 Biologicals and Biosimilars

against that target. The most powerful phenotypic screens do not go into the game with any preconceived notion. It is just: show me the phenotype and I will get the target afterwards. That is how you discover new things. So if that is what you are

alluding to, you are correct. Once you with certain exceptions like neurons, but in

general, if you go after a known target, you get a known antibody.

Markus Gr¨utter: We don’t run genome-wide test of binding but what we do, for example in membrane proteins, is taking a set/a number of diﬀerent proteins as a control experiment, but these are limited and not done at the level that you were asking. But the interface is such that it is as big and selective as antibodies. Initially, distribution is allowed. distribution

when this system was installed, these questions were addressed on a principle basis.

How selective is this on a panel of many different things? You can do selection on entire cells and get specificity to one particular target, so you can fulfill that 9180 requirement in principle. It has not been done in the in vitro case that I presented

only. No further No only. but, in general, this test can be done. Maybe somebody else can comment on that

also?

Dario Neri: Maybe complementing what you just said. There are experiments published by the group of Hans Lehrach in Berlin and Dolores Cahill in Dublin. vayinstitutes.be/ They prepared protein arrays with something like 4,000 proteins and then addressed PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- the question: If you throw an antibody in a dot blot type of assay, do you find other specificities? Occasionally you do. So you may find another spot or two, but it is

http://www.sol not a general rule. Then you may follow it up and see if the aﬃnity is the same. Typically, these reagents tend to be very speciﬁc and you can prove it on an array

host on on host in the level of thousands of proteins.

Brian Roth: The question didn’t relate so much on antibodies because I get the speciﬁcity there but more with these synthetic protein reagents that we are talking about. How deeply those have been interrogated? The other point is that I am a http://www.worldscientific.com/worldscibooks/10.1142/ big believer in phenotypic screens. I didn’t get into this but we actually do a lot of those and in terms of discovering new pathways; that is really the only way to go.

Richard Lerner: If we wanted to be helpful, I’ll come back to the biosimilars for a minute. We could have that now, because I think we should discuss science. Given the way the FDA works, this kind of body (the Solvay conference participants) could develop the Solvay criteria of similarity. And if it was endorsed by a certain body

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World or collection or whatever, this is what the FDA would work with. What would you

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY accept? Because nothing is perfect, not even analytical works on small molecules, as we all know. You would accept let’s say, an acrylamide gel to start with, then ORD, then D-H exchange, then some sort of phenotypic 2D-NMR, and then maybe even a crystal structure, and of course the mass spectrometry. If you had the Solvay criteria, the 10 Solvay criteria that you have to match in order to be a biosimilar and, if the FDA would accept that, the ﬁeld could go forward. Because given the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 275

Session 5: Discussions 275

fact that these are going to be multibillion dollars entities if they are accepted, the people who are trying to put them on the marketplace, would spring for the cash, to build strong analytical capability to match the Solvay criteria. But somebody

has got to sit down and learn to really understand analytical chemistry that applies

to protein and say “alright” according to the scientiﬁc principles of today, these ten (or 7.5) things are acceptable as a biosimilar. Nobody, no agency, no group has ever sat down and done it.

Hiroaki Suga: I want to respond to your question for the “outside of the target”. Because I don’t have X-ray structure I didn’t show today, but we published an AKT2 distribution is allowed. distribution

inhibitor, and we are now writing a paper on the cMET binders. We collaborated

with a MS specialist, his name is T. Natsume. We attached a ﬂag tag and pull it down. He mapped out the whole entire thing. He really found only the target 9180 to which it was targeted. He was very surprised — this could be better than an

only. No further No only. antibody. I think it is the peptide we selected, generally very speciﬁc to the target

we isolated. To the question from the chair about collaborations and how to make available the technologies: I don’t collaborate with industry people at all, because I have a biotech company that will collaborate with the industry. But I am very open to vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS collaborate with any of the academic people and try to work on some new target

- together, or some diﬃcult target that people have failed to identify an inhibitor or whatever. http://www.sol Brian Roth: Let me respond to that. I don’t want to sound like a broken record

host on on host here, but to just put this idea in your mind. I believe what you say if you do “pull down”, you only see the target. But I would point out that if the target is a GPCR for instance, you are probably not going to see it (by MS). These targets are present at trace amounts, they frequently do not ﬂy in a mass spectrometer, you are not going to see them, and those can be the ones that really can cause you problems in http://www.worldscientific.com/worldscibooks/10.1142/ a big way downstream, I guess it is in human populations. There are other ways to proﬁle these compounds, you might think about this.

Markus Grütter: Just to follow up on the question of selectivity. You said that you believe that the antibodies have the selectivity and these scaffolds don’t. There is, from a protein chemistry-based view of these molecules, no reason why these scaffolds shouldn’t acquire the same type of selectivity, in principle. Whether all ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World the experiments are always done is another question, but there is no principal barrier

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY for these scaﬀolds to achieve similar selectivity as antibodies.

Sachdev Sidhu: I would agree with that. I love antibodies as much as anybody else. Scaﬀolds as long as they are designed correctly, it is biophysics, they can work. The problem is that if they are not designed obviously there: they become hydrophobic and they can’t! October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 276

276 Biologicals and Biosimilars

But it is not kind of the point, if you really have to adopt the technology to do what you want to do, then it becomes a fantastic technology. We did that with ubiquitin, these things won’t work for everything but they work very well for that

subset. And if you think about it, in biophysical terms and then design the scaﬀold,

there are plenty of examples where you get the same aﬃnity and speciﬁcity, once the design takes the biology and the biophysics into account.

Kaspar Locher: I have a question for Dr. Suga about scaﬀolds. He made a statement and showed it on his slides that most of the scaﬀolds and their compounds are membrane permeable. (H. Suga: No!) So the questions is actually: would that be distribution is allowed. distribution

something that is good or bad? Is that tunable or not?

Hiroaki Suga: It is tunable but it requires quite a bit of eﬀort. For instance,

9180 what I showed you today was a N-methylated backbone that improved membrane permeability. We had to scan them. We scanned every single residue of the N- only. No further No only. methyl group to identify which one retained activity and then we combine them together. So it requires a bit of eﬀort, like very simple medicinal chemistry. If you have the X-ray structure, you can actually down size it a little bit, which helps a lot too. But whatever you selected, it is not necessarily membrane permeable. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Jason Chin: I have a question for Sachdev Sidhu about the antibiome project. I am curious — you are not going to have infinitely specific reagents. So my question is: at what point are you happy that these are good enough? It is great what http://www.sol you are doing in characterizing them. And actually I think having resources where antibodies are properly characterized is fantastic. But how do you decide that there host on on host is a hundred-fold specificity?

Sachdev Sidhu: This is very simple. You have to start with a user or application in mind. It is the case with any antibody. Nobody designs it to a speciﬁc level. You

http://www.worldscientific.com/worldscibooks/10.1142/ design it to a level where it becomes useful. What we mean by that is; there is no point in launching an antibiome without a very strong network of users. I don’t really care about how tight or speciﬁc it is, as long as the cell biologist who wants touseitsays“itisworking”.ThesameistrueforHerceptinonceitbecameusable, it is ok. If you go back, it might not be 100% speciﬁc. You let the application decide that. We try now to convince the NIH and others to engage in large-scale projects

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World to produce 10,000 antibodies. You set up networks for you know somebody is going

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY to test this immediately and you stop when it is good enough because you want to go on the next one. But otherwise, you get with this incredible mess where you may make it too good, that is a waste of money, and if you don’t make it good enough, nobody picks it up. So it is deﬁned strictly by what needs to be done with it. A 100 nanomolar binder can be great if a cell biologist needs to do an IP (immunoprecipitation) and then he can put it in the micromolar. And that is ﬁne October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 277

Session 5: Discussions 277

if what cell biologists say is: “It’s working!” and then you move on to the next one. That has to be user-driven.

Don Hilvert: I wanted to follow on Brian Roth comment about the synthetic systems that Hiroaki Suga and Dario Neri presented. What is the optimal set-up building

blocks?Doyourestrictyourselvestoavery small subset of chemical space or is everything open? Because if you have a 10 amino acid peptide, and you use 20 building blocks, you have a number of possible combinations that exceeds what you can typically explore experimentally.

Dario Neri: I guess you are referring to DNA-encoding chemical libraries and, even distribution is allowed. distribution

though the concept has been around for many years, there is still work in progress

and there is room for optimization. In my experience, also looking at what other

9180 groups do and activity in our lab, the chemistry that is used nowadays becomes problematic when you go beyond 2-3 sets of building blocks. This coincides also with only. No further No only. the size: getting bigger than 600 Da, which for medicinal chemistry applications is not good. I would claim, but it is a personal opinion, that the technology at this moment is particularly useful to make good quality libraries based on two sets of building blocks with about 1 million members per library. That is where the vayinstitutes.be/ technology works very robustly. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- As you know I organize every two years a congress on this particular topic and I have seen the ﬁeld evolving from companies like GSK constructing libraries with

http://www.sol billions of molecules with 4 sets of building blocks of very poor quality — going down to libraries of 2-3 sets of building blocks resulting in 1 to 10 million compounds but

host on on host with better quality and performance. I think a lot of development is still possible. Trying to make these molecules more compact, and smarter in a way.

Hiroaki Suga: In my case, you can put anything, any building block, you can change the side chain of amino acids and stuﬀ like that. But I am not so interested in it http://www.worldscientific.com/worldscibooks/10.1142/ and I think it is not really important. I think the diversity of the side chains of amino acids is good enough already. There is more to be gained by changing the “backbone”. We are making more eﬀorts with N-methylated backbones or — which I haven’t talked about today — we could have oxazole or thiazole kinds of backbones where you are missing the hydrogen donor from the peptide backbone. And that is more important to the development of drug-like molecule in the future.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Chris Walsh: I would like to go back to the interface of small and large molecules

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY with this question. In some sense it is applied rather than basic science, but there is of course tremendous interest in antibody-drug conjugates (ADC) at least in the cancer area. One of the two approved drugs, the one from Seattle Genetics has a peptide as the toxin. I know Dr. Suga wants to get away from antibodies, but at some level one might say: appending some of your optimized peptide scaﬀolds as antibody-drug conjugates might be interesting. I would like the panel’s view October 15, 2014 17:11 WSPC Proceedings - 9.75in x 6.5in master/pg 278

278 Biologicals and Biosimilars

on this: Should we really be excited about ADCs for cancer? Otherwise are they just a fad because very little has been said about them here. I understand one side brought them up. We are not about applications but at the interface of small

molecules and biologicals, I would be interested. We heard seven presentations, we

didn’t hear anything about ADCs. Yet in a diﬀerent context, that is all the big companies would be talking about. Whether it is Ambrex technology or something else. Should we not think that it is an important scientiﬁc advance? Is it the future of cancer antibody therapy?

Richard Lerner: I think you are absolutely right. I have been around long enough distribution is allowed. distribution

to watch the evolution of this. Immunotoxins were the original armed antibody

right? People fooled around with it. And the reason is that antibodies in general are not very toxic. They have an activator domain and so forth. If you look at 9180 most of the multibillion dollar antibodies that are out there, there is a question

only. No further No only. about these. I believe that the issue you raised is very important for the future.

However, if you look at Seattle Genetics, the total chemical armamentarium of the synthetic world has not until now been focused, and there is tremendous room for making very intelligent linkers, tags, and so forth. . . I think we are going to see some heavy-duty synthetic guys, getting into this business. More payloads, more vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS linkers, smart linkers and so forth. . . Seattle Genetics is very successful, and rightly

- so, but for a while they were the only game in town.

http://www.sol Chris Walsh: There are three pieces to antibody-dependent conjugates: one is the antibody. Should we also be thinking about nanobodies or peptide fragments? The

host on on host second is the linker technology - for better or worse (the third piece is the drug). I started Immunogen in 1981 but this has been a very long time coming. Also the small molecules had to be suﬃciently potent. They had to be picomolar. They are very few, but I would propose Dr. Suga’s compounds might get to that point. I don’t hear anybody talking about using optimized peptides. Both of the ADC that http://www.worldscientific.com/worldscibooks/10.1142/ has been approved target microtubules in sort of an old fashion approach.

Richard Lerner: So the first armed antibody was actually taken off the market because it was too hot. So you have to tune the affinity constant of the antibody, with the hotness so to speak of the organic compound. Because if, as is the case of this compound, a single molecule kills the cell, you can’t use it.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Dario Neri: I think the discussion is an interesting one. There are I think oppor-

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY tunities for improvement. For example, the ﬁrst antibody that was on the market in the US (not in Europe and was later taken oﬀ the market) was very dirty, like 40% of the product was not even conjugated to the payload. There are some payloads, which are very interesting because they show the direction for the future. Because whatever you inject, as an ADC, at some stage, it is going to get through a clearance organ, typically the liver. So, 100% of what you inject at some stage October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 279

Session 5: Discussions 279

targets the liver. Maybe it also targets the tumor to a minute extent, but largely the liver. I think we will see developments in self-detoxifying drugs. If you think about maytansine structures, there are esters, epoxides and you can think of course

of molecules, which get cleaved and detoxiﬁed as they get out of the body. So I

think it is an area where modularity will tell us if we are good enough to get better targeting agents, better linkers and I would claim self-detoxifying payloads.

Arne Skerra: I foresee a bright future for drug conjugates once we overcome the dogma of antibodies to be the preferred carriers. And there are two very simple reasons. One is the liver uptake of antibodies, which actually killed Mylotarg (from distribution is allowed. distribution

Pﬁzer), the product we were talking about. And the other reason is poor tissue

penetration and extravasation (we have also talked about in the session). So if we go to either stable small antibody fragments or alternative scaﬀolds, then we will 9180 be able to create a totally new generation of much improved drug conjugates. only. No further No only. Richard Lerner: What is the basis of that statement?

Arne Skerra: I gave two arguments. One is for example liver toxicity, which is the limiting eﬀect (at least it was for Mylotarg) for many preclinical models, as I

vayinstitutes.be/ heard from the biopharmaceutical industry. And we all know there are receptors in PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- the liver, like the asialoglycoprotein receptors, which take up the antibodies after a certain time of circulation. This intoxicates the liver. Once we can avoid that, we can improve the ratio of tumor toxicity to organ toxicity. http://www.sol

Richard Lerner: So if you engineer out the scFv binding piece of an antibody? host on on host Arne Skerra: We are in the process of doing that: we have a collaboration already and it is quite promising.

Sachdev Sidhu: In my opinion it is kind of a middle ground where clearly antibodies http://www.worldscientific.com/worldscibooks/10.1142/ have to be the bench mark for ADCs and then you have to look at them and decide whether a particular alternative gives you a signiﬁcant advantage in one of the three key areas, I would say: clearance, penetration but, as important, internalization and release. And that is a science problem. You have to ask yourself: Can I get something which is designed to either internalize or release better? But you have to set it up against the benchmark, which is the antibody, and rationally approach that. I think there is a large opportunity there if you can ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World mechanistically understand how to get something in and then the molecule out, but NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY you have to do it in relation to the antibody. I don’t think you can do it without a bench marking process.

Chris Walsh: Prof. Suga, any comment?

Hiroaki Suga: Surprisingly, the pharmaceutical companies did come back to think about those. Once they see our peptides. . . For instance, they can bind extremely October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 280

280 Biologicals and Biosimilars

well to the target protein but do not inhibit interactions for example with growth factors. And these are really good targets to conjugate with a drug and bring it to the desired target. I think they become very much interested in peptides.

Taking advantage of the small microcyclic peptides; they are very easy to get

conjugated, and the puriﬁcation is also easy, so that may be the bible for that kind of concept, i.e., peptide-conjugates can be potent enough to target the desired cancer.

Arne Skerra: I have a technical question to Dr. Suga. I was amazed of how eﬃciently you can incorporate methylated amino acids — or even D-amino acids — into your distribution is allowed. distribution

peptide. At which point does the mechanism of the ribosome limit your peptide

chemistry?

9180 Hiroaki Suga: We have studied that a lot, we have many papers (like 5) on that. The D-amino acids can be incorporated in the initiation extremely eﬃciently, no matter only. No further No only. what the side chains are. It is during elongation that we get problems. Single amino acids or D-amino acids can be inserted at certain side chains, but not many of them. But the major problem is actually consecutive incorporation, which is completely inhibited. That is why we usually have only a single D-amino acid or having it at vayinstitutes.be/ a speciﬁc position where it is ok with respect to the elongation position. On the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- other hand, some N-methylated amino acid, like charged N-methylated amino acid (due to charge eﬀects), are very poorly incorporated at the elongation stage. We

http://www.sol have all knowledge of what you can do best and what you cannot do well, and we designed a genetic code in that way. host on on host Gebhard Schertler: A question for the future. We can differentiate cells and do a lot with selection. And selection has make such progress. Are we fast enough to take cells from a patient and deliver an agent that is specific for that patient? That is a first question. My second question is: in cancer for example, the main thing is not http://www.worldscientific.com/worldscibooks/10.1142/ to kill mainly the cancer cells, a few months later they come back. Are we having a homogeneous cancer population or is there something like there are cells that are just becoming the cancer cells (people have been talking about cancer stem cells)? What is your opinion about that? Can it be solved just by using two targets at the same time? But really the main thing is still that, even if you kill many cells, we can see that it often comes back. We misunderstand how metastases form, how it happens. What is the research in there? What can selection do about that? ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Richard Lerner: You are talking about custom therapy. Do you have time to make custom therapy for individuals in the antibody world? The answer is yes in some situations and no in others. In the case of lymphomas, every lymphoma has a different antibody on the surface. So you could in principle, take a sample of the lymphoma at the time of the first biopsy, figure out what the antibody is on the surface and make an armed antibody or killer antibody, custom therapy and charge less and less than the custom dendritic ‘company’ charges. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 281

Session 5: Discussions 281

There is one custom therapy immunology, publicly traded company, that is Dendreon. They take your tumor, they build cells that don’t like your tumor and put these back in the humans. That is not very eﬀective. The problem is that if

you do not pick a target that the cells requires, what happens is the cells will dump

the target and then go on and metastasize and do all kinds of bad stuff. If you are going to do that drill, and I think it is possible in the future, if the payers pay for it, you have to find a target that the cells can’t get rid of. And to find a target that is (A) a tumor marker and (B) the cells cannot get rid of it, then you have a chance to do exactly what you say. But a tremendous amount of effort is done on killer antibodies so to speak, against targets that can be missed by the tumor cells. distribution is allowed. distribution Of course, the cells want to survive, the population dumps that and that is the end

of it. (Schertler speaking about the fact that many cancers come back) For the second part of the question, of course the Holy Grail is to get the cells to 9180 differentiate again, because the phenotype of cancer is dedifferentiation. If you had only. No further No only. an agonist that would cause the dedifferentiated cell to redifferentiate, you would probably stop the cancer. A lot of cancers make the phenotype of the cells from which they come. For example, if you have an insulinoma in a patient and they are making brown fat (cachexia) by some mechanism, the tumor is hot, because you

vayinstitutes.be/ are making heat, not ATP. If you could make that tumor cold by dediﬀerentiation, PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- it probably would not be a tumor anymore, it would be a cell. There are precious few tumors that have the mature phenotype of the parental system. In other words: what the oncologists say is that sometimes you can recognize where they came from, http://www.sol so if you could, with dedifferentiation, make them go back to where they came from, you have got something really nice. host on on host Sachdev Sidhu: I think that is a great point. We are really much thinking to kill cancer but perhaps what you need to do is remove most of it and kind of achieve “peace with honors”. . . And that is better done with agonist rather than antagonist. The cancer field is very heavily focused on antagonists and signal blocking. What http://www.worldscientific.com/worldscibooks/10.1142/ is needed and is possible is to combine that with turning on differentiation signals.

Markus Aebi: We use penicillin since about 60 years. Now penicillin is no longer in use because there is resistance and Fungi have been using penicillin for 50 million years and it is still in use that we did not read the description how to use penicillin. We know the answer — we probably have to target multiple points in a cell, in a pathway. And then I wonder: Are there strategies, using combinatorial methods ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World — as you are using very nicely — to identify those target pairs or target triplets, NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY to make therapy much more eﬃcient? We know from HIV treatment that this is the solution. You would use combinatorial methods to ﬁnd the magic bullet but

we know (at the same time) that this magic bullet does not exist because you are targeting evolving systems. We will never reach that goal.

Richard Lerner: That is exactly right. And so Christopher Walsh raised one Holy Grail, which does not address that point, and is making antibodies more toxic. The October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 282

282 Biologicals and Biosimilars

other thing that is a big deal in the field is to make bispecific antibodies that target two targets at once. Every big pharma and half the small pharma in the antibody space are making bispecifics. The nice thing about infectious antibody libraries is

that you obtain bispeciﬁcs whether you want it or not, automatically you have a

bispeciﬁc. And bispeciﬁcs would help get around the problem that you raised.

Markus Aebi: It might be if you go for a single molecule with two functions. I would rather love to see two different targets, with combinatorial methods, or I did not understand well your answer. . . (Richard Lerner interrupts to define what he means with bispecific: one antibody for two targets, with one head having the distribution is allowed. distribution

target A, and the other head targets B).

OK great! If you can come up with that system, that is fantastic. I think the chance is much higher to ﬁnd two diﬀerent targets and two individual antibodies, if 9180 you want to stay with antibodies. I am not an expert, but are there combinatorial

only. No further No only. methods then to get that?

Richard Lerner: Yes. Everybody has its own way of making such a system. Genen- tech is using knobs and holes. We use natural recombination at the protein level and so on and so forth. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Dario Neri: Briefly to integrate the answer. Of course, the history of cancer therapy is history of combination therapy, which has been very efficacious in certain types of leukemia and lymphoma but actually not very efficacious in the field of solid tumors. http://www.sol There are a few types in which chemotherapy was actually shown to eradicate cancer metastases of solid tumors but usually this doesn’t work. I think we often forget that host on on host in order to try combination in the clinic, you need two agents that have typically gone into the clinic separately, show to have single agent activity, and maybe 5-6 years down the road, you can try the combination. This is actually limiting progress alot. http://www.worldscientific.com/worldscibooks/10.1142/ At the same time, I think it is fair to say, If you looked at the last ASCI meeting, probably the biggest excitement was again from combination of two antibodies: anti- TLR4 and anti-PD1, where all of a sudden you see a response rate in metastatic melanoma, which is a deadly disease, and which has never really been reported before. So I think the direction is clearly that one. In the mouse, it is very easy to show cure with combination. What I have learned is that translation is very slow and we should maybe think of how to make it more efficient. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Markus Aebi: This is exactly what I was asking for. Are there tools to speed that up, you cannot do that in patients or in mice, it takes simply too long. Are there combinatorial methods in this respect?

Chris Walsh: This is really a follow up to an earlier question and Richard Lerner and his response around stem cells. You showed this amazing ability to ﬁnd antibodies, which should diﬀerentiate CD34+ cells, why wouldn’t you now go to the epithelium October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 283

Session 5: Discussions 283

mesenchymal transition of cancer stem cells from different origins? However controversial that hypothesis is about whether they really drive metastasis, and ask can you readily find agonists antibodies which would work on different kinds of stem

cells to block that transition?

Richard Lerner: That is right on, we are doing exactly that. The question is: What is the selection? And the selection we are using is: there is a breast cancer model in mice where you put human breast cancer tumors cells into the footpad and they all metastasize to the lung. Our selection is: you infect them with the antibody library and pull out those that do not metastasize. And then you do it again. And distribution is allowed. distribution

our guess is that those would be cells that cannot metastasize.

Lode Wyns: When I listen to the whole discussion it is obviously biological and

9180 medical science. But I think there are many other ﬁelds. I see many applications and we didn’t say a word about it: agriculture and plant science for example that only. No further No only. was not touched in this meeting. There are a few botanists here. And then the other aspect (what Markus Aebi tried to raise a moment ago) is: to what extent do we want to provide these tools to the community? I am very positive about the idea of the anti-ome. That would be extremely eﬃcient in many vayinstitutes.be/ domains of science. On the other hand I think it will remain very individual in the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- future in a sense that if you want to study mechanisms, such as ﬁbrillogenesis or if you are looking at GPCRs, then one antibody for one target is not enough. But we

http://www.sol see that if we have ten or more antibodies, they each have their peculiar properties and use them for the fundamental aspects of the process. host on on host Sachdev Sidhu: A comment on that. I think you made a great point. The interesting thing is that medicine has to be conservative and it might be that new frameworks, new ideas you can actually test a lot easier in agriculture or in industry where you are not worried about immunogenicity in patients. I think that it is a good area http://www.worldscientific.com/worldscibooks/10.1142/ to get into, where new ideas can be explored with industrial applications that can then maybe make their way into medicine. And the whole idea of distributing that I think is critical because they (antibodies) have to be used to ﬁgure out a funding structure and an impact where you can actually be doing something beyond protein engineering. Our hypothesis has been to “Act locally and distribute globally” and what I mean by that is: you build a network around a center and work on problems where you know you have to cross ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World the wall/street, and use those to ﬁgure out what is working and then be able to

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY distribute those to anybody else in the world who wants them. Otherwise it doesn’t work. If you try to produce a lot of stuﬀ that people don’t even know about, you simply end up with freezers full of these things.

Richard Lerner: But Dev Sidhu you make one exception. I think you have been wildly successful and unusual. One thing you have to realize is that the world who October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 284

284 Biologicals and Biosimilars

wants an antibody is not as sophisticated a molecular biologist, as you are. If you want this to get attraction, you cannot only send the gene. You need to send them milligrams of the protein and the condition is if they want more, then “Here is the

gene and you make more”. But all these enterprises have failed because the world

has many genes around and people don’t know how to use how to express them. Send a milligram and then you will see traction.

Gebhard Schertler: Can I ask a question to Dario Neri? We have often toxicity when you want to deliver a toxin or something to a target. We have been very good at starting to get complementing systems for assays, where we actually make the distribution is allowed. distribution

proteins with two things come together. How far are we in payloads that assemble

on the tumor cell?

9180 Dario Neri: I think this the Holy Grail of protein engineering. How to engineer toxicity as a result of binding? To a certain extent this is what antibodies do. only. No further No only. They swim in the body as naked IgG — they don’t do anything but, if they are on the target cell, then they interact with the Fcγ receptors and do the job if accessory cells are around. That is the way we are engineering toxicity after binding. From my perspective, it is a very clear ﬁeld for ingenuity and development but, on vayinstitutes.be/ the other hand, we always have to ask if they are good to reach the target, and PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- this is less trivial than one thinks. Typically, companies develop products without imaging, without really doing biodistribution studies in humans. That is going

http://www.sol blind, that is assuming that the product goes to the target but maybe it does not. There is a beautiful paper of Genentech a few years ago in which they gave

host on on host Herceptin (Trastuzumab monoclonal antibody) at very high dose to tumor bearing mice. They looked at the section of the tumor and you would have expected that every single tumor cell should be covered by the antibodies but actually only few perivascular tumor cells were reached. Again highlighting the fact that antibodies don’t extravasate so well and the few that do extravasate ﬁnd the antigen, bind http://www.worldscientific.com/worldscibooks/10.1142/ and stay there. So even simple concepts, like reaching a target in vivo, are not that trivial.

Richard Lerner: There are a lot more important points that are not raised. Some- times you get too much antibodies. Because, for example, if you take the “brown fat” antibody, it will work at 0.1 mg/ml. but if you go up to 1 mg/ml like your small molecules, then you suppress the system. So it could be in the case of Herceptin ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World that you could actually give less, it might not downregulate the receptor and be

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY better. But companies don’t want to give less because to people you’ve been giving less, you can only charge less.

Markus Gr¨utter: I have to close the discussion session. A lot of questions are still open and a lot needs to be done but there are great opportunities with these molecules. I thank all the panel members. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 285

Session 6 Proteins in Supramolecular Machines distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 286

distribution is allowed. distribution

Cryo-EM density map of the 80S ribosome from Trypanosoma brucei ˚ 9180 at 5 A resolution, reconstructed from over 200,000 particle projections. Left: front view, looking from the factor binding site into only. No further No only. the intersubunit corridor, along the path of the tRNA. Right: back view. Ribosomal RNA expansion segments are rendered in diﬀerent colors, and E-site tRNA is labeled as “E-tRNA.” vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS Modiﬁcation of ﬁgure from:

- “High-resolution cryo-electron microscopy structure of the Trypanosoma brucei ribosome”, Y.Hashem,A.desGeorges,J.Fu,S.N.Buss,F.Jossinet,A.Jobe,Q.Zhang,H.Y.Liao, R.A.Grassucci,C.Bajaj,E.Westhof,S.Madison-Antenucci,J.Frank,Nature 494, 385 http://www.sol (2013); Copyright held by Nature Publishing Corp. host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 287

287

ASSEMBLY OF FILAMENTOUS TYPE 1 PILI FROM UROPATHOGENIC ESCHERICHIA COLI STRAINS

RUDOLF GLOCKSHUBER Department of Biology, ETH Z¨urich, Institute of Molecular Biology and Biophysics Schafmattstrasse 20, CH-8093 Z¨urich, Switzerland distribution is allowed. distribution Understanding the assembly mechanism of supramolecular complexes and their

molecular architecture at atomic resolution is one of the major challenges in biol-

9180 ogy. Despite the increasing number of solved three dimensional structures of large supramolecular assemblies and a large body of data on cellular factors required for only. No further No only. their assembly in vivo, it is still extremely challenging to reconstitute supramolecular assemblies from all puriﬁed components in vitro and to obtain a complete picture of the assembly process that integrates structural, biochemical and in vivo data. In the following, I will summarize recent results on one of the very few vayinstitutes.be/ supramolecular assemblies that could be successfully reconstituted in vitro,ofwhich PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- all structural components and assembly factors are known, and where in vitro and in vivo data result in a uniform picture of the assembly process. The results are in line with Richard Feynman’s famous statement “What I cannot create, I do not un- http://www.sol derstand”, in particular, because the reconstitution of the entire assembly process

host on on host in vitro led to the discovery of a previously unknown family of protein folding catalysts and the identiﬁcation of a membrane catalyst that accelerates the assembly of protein subunits independently of energy.

Assembly of ﬁlamentous type 1 pili from uropathogenic http://www.worldscientific.com/worldscibooks/10.1142/ Escherichia coli strains Type 1 pili of uropathogenic Escherichia coli strains are ﬁlamentous, 0.5–2 μmlong surface organelles anchored to the outer bacterial membrane and are required for bacterial attachment host cells. They are composed of the four structural subunits FimH, FimG, FimF and FimA [1] (Fig. 1). The main structural subunit FimA assembles into a right-handed, helical quaternary structure and forms the pilus rod.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World FimH, FimG and FimF associate to a short tip fibrillum composed of a single copy of FimH at the tip of the fibrillum and one or several copies of FimF and FimG. The NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY adhesin FimH is the first subunit that is incorporated into the pilus. It is composed of an N-terminal lectin domain and a C-terminal pilin domain that interacts with

the following subunit FimG. The FimH lectin domain recognizes the mannose units of the glycoprotein receptor uroplacin 1a of epithelium cells of the urinary tract [2]. FimA, FimF, FimG and the pilin domain of FimH are structurally homologous proteins that share an immunoglobulin-like fold lacking the C-terminal β-strand October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 288

288 R. Glockshuber

[3]. In addition, FimA, FimF and FimG possess an N-terminal extension of 15– 20 residues, termed donor strand, which inserts into the preceding subunit in an anti-parallel orientation relative to its C-terminal β-strand and thereby completes

its fold. This donor strand complementation mechanism is the common structural

principle underlying quaternary structure formation and explains why the adhesin FimH can only be incorporated at the tip of the pilus [3]. Type 1 pili are assembled via the chaperone-usher pathway [4] (Fig. 1). This term refers to the fact that the periplasmic chaperone FimC and the assembly platform (usher) FimD in the outer membrane are essential for pilus biogenesis. FimC interacts with all pilus subunits in the periplasm and delivers them to FimD at the distribution is allowed. distribution outer membrane. FimC-subunit interactions diﬀer from subunit-subunit interac-

tions in that an extended polypeptide segment of FimC is inserted in the opposite (parallel) orientation into the pilin fold of the subunit. FimD recognizes incoming 9180 FimC-subunit complexes, catalyzes subunit assembly (see below) and mediates the only. No further No only. translocation of folded subunits to the extracellular space. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

Fig. 1. Subunit composition and biogenesis of type 1 pili from Escherichia coli. (Top) Electron micrograph of the E. coli K12 wild type strain W3110 bearing type 1 pili. (Bottom) Mechanism of DsbA-, FimC- and FimD-catalyzed assembly of type 1 pili, and quaternary structure of the pilus rod and the tip ﬁbrillum (see text for details). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 289

Assembly of Filamentous Type 1 Pili from Uropathogenic Escherichia Coli Strains 289

Type 1 pili are extraordinarily stable against unfolding and dissociation. To study the origin of this extraordinary stability, we constructed subunit variants that lacked the N-terminal donor strand and were artiﬁcially extended at the C-terminus

by the donor strand of a neighboring subunit [5]. This enforced intramolecular,

anti-parallel self-complementation and thus mimicked the conformational state of subunits in the pilus. All self-complemented variants proved to be extremely stable, with free energies of unfolding in the range of 67–85 kJ mol−1 [5]. In addition, they did not attain their folding equilibrium at physiological temperature due to extremely high activation energy barriers of unfolding and refolding, explaining the almost inﬁnite stability of pili against dissociation [5, 6]. Surprisingly, bimolecu- distribution is allowed. distribution lar complexes between subunits and synthetic donor strand peptides show similar,

high stabilities. An impressive example is the complex between FimG and the donor strand peptide of FimF (DsF). It is the most stable, non-covalent protein- 9180 ligand complex known to date, with a dissociation constant of 1.5 10−20 Mandan only. No further No only. extrapolated dissociation half-live of 4 billion years [5, 7]. This opens numerous possibilities for technical applications, e.g., the use of the DsF peptide as tag in aﬃnity chromatography or for the stable immobilization of proteins on surfaces [7]. All pilus subunits exhibit very low rates of spontaneous folding. Speciﬁcally,

vayinstitutes.be/ the main subunit FimA shows a folding half-life of 1.6 hours [8]. We identiﬁed the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- chaperone FimC as highly efficient and specific catalyst of subunit folding [8, 9]. The rate-limiting step in the catalysis is the binding of the unfolded subunit to FimC, while subunit folding on the surface of FimC occurs with half-lives of less http://www.sol than a second. FimC thus can accelerate folding of subunits by more than 104-fold [8]. host on on host All structural pilus subunits possess an invariant, structural disulfide bond, which is introduced into newly translocated subunits by the dithiol oxidase DsbA (Fig. 1). We found that FimC only recognized unfolded pilus subunits in which the disulfide bond was already formed [10]. As only FimC-bound subunits are assem-

http://www.worldscientific.com/worldscibooks/10.1142/ bly competent, this mechanism guarantees that only disulﬁde-intact subunits are incorporated into the pilus [10]. The kinetic analysis of FimC- and DsbA-catalyzed folding of FimA, together with the in vivo concentrations of both catalysts in the periplasm, predicted and in vivo half-life of FimA folding of approximately two seconds [10]. Thus, FimC and DsbA very eﬃciently overcome the kinetic bottleneck of the slow, spontaneous folding of pilus subunits. The most fascinating component in the type 1 pilus assembly system is certainly

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World the assembly platform FimD. It is composed of an N-terminal domain (FimDN)that β NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY recognizes incoming FimC-subunit complexes [11], a -barrel transmembrane domain forming the pore for subunit translocation, a plug domain and a C-terminal periplasmic domain [12, 13]. FimD exists in two distinct conformations: in its inactive state, the plug domain occupies the translocation pore. The reconstitution of pilus assembly from FimC-subunit complexes in the presence of catalytic amounts of FimD [14] showed that FimD can only be converted to its active state by the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 290

290 R. Glockshuber

FimH-FimC complex. The solved X-ray structures of both conformations demonstrated that the FimH lectin domain occupies the translocation pore in the ternary FimD-FimC-FimH complex and displaces the plug domain towards the periplas-

mic side of the membrane [12]. FimA polymers only form slowly from FimC-FimA ◦

complexes, with a half-life of about 10 hours at 37 C [14]. FimD proved to be a very eﬃcient catalyst of FimA polymerization, and pilus assembly was completed within 20 minutes in the presence of catalytic amounts of FimD (0.5 equivalents) [14]. FimD-catalyzed pilus assembly functions independently of energy and is exclusively driven by the higher stability of subunit-subunit complexes compared to FimC-subunit complexes [14]. The catalytic cycle of FimD starts from a state in distribution is allowed. distribution which the last subunit at the growing end of the pilus is capped by FimC. The next

FimC-subunit complex then binds to the N-terminal FimD domain (FimDN). The donor strand of the incoming subunit then displaces the FimC molecule capping the 9180 pilus end. The last step of the cycle is a shift of the pilus by the length of a subunit only. No further No only. towards the extracellular space, and the chaperone bound to the newly incorporated subunit caps the pilus [14]. A detailed analysis of the individual rate constants of the binding of FimC-subunits to FimD and the rates of subunit-subunit contact formation revealed that FimD determines the speciﬁcity of the assembly process

vayinstitutes.be/ in that contacts between natural pairs of subunits are formed dramatically faster PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- than wrong subunit-subunit contacts [14, 15]. The successful reconstitution of pilus assembly should allow a quantitative prediction of the length distribution of pilus rods formed in the presence of FimD. http://www.sol

host on on host Development of antiadhesive drugs for treatment of urinary tract infections The high frequency of urinary tract infections (UTIs), with about 50% of all women suﬀering from at least one UTI during their life [16], and the high rates of UTI re-

http://www.worldscientific.com/worldscibooks/10.1142/ currence after therapy with antibiotics, require new strategies for UTI treatment. Recurrent UTIs are most likely a consequence of intracellular bacterial communities that persist in epithelium cells of the urinary tract. As internalization of the bacteria by epithelium cells is type 1 pilus dependent, an obvious strategy to prevent recurrent UTIs is the development of antiadhesive drugs. Towards this goal, two diﬀerent concepts have been followed. First, synthetic inhibitors of pilus assembly termed pilicides have been identiﬁed that inhibit de novo pilus assembly by binding

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World to the assembly chaperone [18]. Second, a series of α-D-mannosides that mimick the natural glycan of the target receptor and bind to the isolated FimH lectin domain NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY in the low nanomolar range has been developed [19]. The advantage of these FimH antagonists is a competitive inhibition of tissue adhesion of pili that have already

formed. Both types of antiadhesive drugs may be used in the future in combination with classical antibiotics. An intriguing property of type 1 piliated E. coli strains is their ability to increase their aﬃnity to target surfaces under tensile mechanical force [20], e.g., under ﬂow October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 291

Assembly of Filamentous Type 1 Pili from Uropathogenic Escherichia Coli Strains 291

conditions during excretion of the urine. This “catch bond” behavior most likely increases the life time of tissue-bound bacteria and the probability of pathogen uptake by urinary epithelium cells, thus promoting bacterial persistence inside host cells

during therapy with antibiotics. The “catch bond” behavior has been attributed

to two different conformational states of FimH observed in the three-dimensional structures of ligand-free FimH in the context of the tip fibrillum and the isolated FimH lectin domain with bound ligands [20]. While the FimH lectin domain in the tip fibrillum shows specific interactions with the FimH pilin domain and an open conformation of its carbohydrate binding pocket, the isolated lectin domain exhibits a more elongated conformation with a narrower binding pocket and strong distribution is allowed. distribution conformational changes in the surface loops that interact with the pilin domain in

full-length FimH. These data, together with the observation that the isolated lectin domain shows an about 100-fold higher affinity for carbohydrate ligands compared 9180 to full-length FimH in the tip fibrillum, have led to the hypothesis that ligand bind- only. No further No only. ing of FimH favors the separation between the lectin and the pilin domain and that, conversely, force-induced domain separation increases the affinity for ligands. This suggests that full-length FimH, rather than the isolated FimH lectin domain, is the physiologically relevant drug target. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Acknowledgments This project was supported by the ETH Z¨urich and the Swiss National Science Foundation (SNF) within the framework of the NCCR Structural Biology program http://www.sol and by SNF grants 310030B-138657 and 31003A-122095 to R.G. host on on host References 1. E. Hahn, P. Wild, U. Hermanns, P. Sebbel, R. Glockshuber et al., J. Mol. Biol. 323, 845 (2002).

http://www.worldscientific.com/worldscibooks/10.1142/ 2. G. Zhou, W.-J. Mo, P. Sebbel, T. A. Neubert, R. Glockshuber et al., J. Cell Sci. 114, 4095 (2001). 3. D. Choudhury, A. Thompson, V. Stojanoﬀ, S. Langermann, J. Pinkner et al., Science 285, 1061 (1999). 4. A. Busch, G. Waksman, Philos. Trans. R. Soc. Lond. B. Biol. Sci. 365, 1112 (2012). 5. C. Puorger, O. Eidam, D. Erilov, M. Vetsch, G. Capitani et al., Structure 16,

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 631 (2008). 6. D. Erilov, C. Puorger, R. Glockshuber, J. Amer. Chem. Soc. 129, 8938 (2007). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 7. C. Giese, F. Zosel, C. Puorger, R. Glockshuber, Angew. Chem. Int. Ed. 51, 4474 (2012).

8. C. Puorger, M. Vetsch, G. Wider, R. Glockshuber, J. Mol. Biol. 412, 520 (2011). 9. M. Vetsch, C. Puorger, T. Spirig, U. Grauschopf, E. U. Weber-Ban et al., Nature 431, 329 (2004). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 292

292 R. Glockshuber

10.M.D.Crespo,C.Puorger,M.A.Sch¨arer,O.Eidam,M.G.Gr¨utter et al., Nat. Chem. Biol. 8, 707 (2012). 11. M. Nishiyama, R. Horst, O. Eidam, T. Herrmann, O. Ignatov et al., EMBO J.

24, 2075 (2005).

12. G. Phan, H. Remaut, T. Wang, W. J. Allen, K. F. Pirker et al., Nature 474, 49 (2011). 13. S. Geibel, E. Procko, S. J. Hultgren, D. Baker, G. Waksman, Nature 496, 243 (2013). 14. M. Nishiyama, T. Ishikawa, H. Rechsteiner, R. Glockshuber, Science 320, 376 (2008). distribution is allowed. distribution 15. M. Nishiyama, R. Glockshuber, J. Mol. Biol. 396, 1 (2010).

16. B. Foxman, P. Brown, Infect. Dis. Clin. North Am. 17, 227 (2003). 17. I. U. Mysorekar, S. J. Hultgren, Proc. Natl. Acad. Sci. USA 103, 14170 (2006). 9180 18. V. Aberg, F. Almquist, Org. Biomol. Chem. 5, 1827 (2007). only. No further No only. 19. M. Scharenberg, O. Schwardt, S. Rabbani, B. Ernst, J. Med. Chem. 55, 9810 (2012). 20. I. Le Trong, P. Aprikian, B. A. Kidd, M. Forero-Shelton, V. Tchesnokova et al., Cell 141, 645 (2010). vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 293

293

HIV ENVELOPE AND INFLUENZA HEMAGGLUTININ FUSION GLYCOPROTEINS AND THE QUEST FOR A UNIVERSAL VACCINE

IAN A. WILSON Department of Integrative Structural and Computational Biology, The Scripps Research

distribution is allowed. distribution Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037, USA

9180 My view of the present state of the structure and function of viral

only. No further No only. fusion glycoproteins, their recognition by the immune system, and

universal vaccines Human immunodeﬁciency virus type 1 (HIV-1) and inﬂuenza viruses remain major threats to human health that plague mankind with continual epidemics and

vayinstitutes.be/ sporadic pandemics. These viruses display surface glycoproteins for binding to re- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- ceptors of the host cell and fusion machinery to enable the viral and cell membranes to fuse so that the viral genome can be deposited into the cell for initiation of viral replication. Blocking of these activities by small molecules and small proteins or by http://www.sol antibodies would neutralize the virus and prevent or ameliorate infection. Although neutralizing antibodies can readily be elicited against either virus, they usually tar- host on on host get regions of the virus envelope with little functional relevance. Viral evolution can then easily outpace antibody adaptation, resulting in strain-speciﬁc antibodies with little or no broadly neutralizing properties. The development of vaccines that elicit broadly neutralizing antibodies as a means for universal protection against

http://www.worldscientific.com/worldscibooks/10.1142/ these two viruses, remains an illusive but important goal. The pathogenesis of HIV-1 and influenza differ significantly: whereas HIV-1 generates a persistent infection, influenza is typically cleared within several days. However, similarities can be drawn in their ability to evade protective immunity and the mechanisms they use to infect cells. The critical first step in viral infection is attachment of the virus to the host cell, followed by fusion of the viral and hosts membranes and viral entry into host cells. In the case of influenza and HIV-1, these complex processes are

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World mediated by their respective trimeric Type 1 membrane fusion glycoproteins, the

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY hemagglutinin (HA) for inﬂuenza and envelope (Env or gp160) for inﬂuenza and HIV-1, respectively.

HA-mediated entry process – The HA trimer is derived from a single polypeptide precursor, HA0, which is proteolysed extracellularly resulting in a metastable mature trimer, composed of two polypeptides, HA1 and HA2. Structurally, the trimer can be subdivided into a membrane-distal “head” and a membrane-proximal October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 294

294 I. A. Wilson

“stem” [1, 2]. HA1 comprises the head and contains the sialic acid receptor-binding site. The HA stem region consists primarily of HA2, in addition to parts of HA1, and composes the fusion machinery, which is triggered in the low pH environment

of late endosomes. Upon acidiﬁcation, the HA trimer undergoes major conforma-

tional rearrangements that directs insertion of its fusion peptide into the endosomal membranes and leads to subsequent fusion of host and virion membranes. Despite decades of work, the details of how these viral glycoproteins trigger have still not been determined nor how to adequately prevent fusion or receptor binding. Prior to 2008, identification of antibodies that broadly neutralize different strains and subtypes of influenza virus were scarce and only one mouse antibody that bound to distribution is allowed. distribution different subtypes was known [3]. Interestingly, in a series of unforeseen discoveries,

three independent groups identiﬁed human antibodies that cross-react with multiple inﬂuenza subtypes [4–6]. As opposed to almost all other known HA antibodies that 9180 typically bind to the head region of HA and likely prevent receptor binding, these only. No further No only. newly discovered antibodies bind the less accessible stem portion of the HA that houses the fusion machinery. Crystallographic and biochemical studies indicated that the antibody neutralizes the virus by blocking conformational rearrangements associated with membrane fusion, thus interfering with the virus’s ability to deliver

vayinstitutes.be/ its genetic material into host cells and initiate infection. These ﬁndings have opened PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- the door to further explorations into blocking the viral fusion apparatus as means to engineer a universal ﬂu vaccine [7, 8].

http://www.sol Env-mediated entry process – The HIV-1 membrane fusion process and subsequent viral entry into the host cell is even less well understood compared to HA-

host on on host mediated entry [9, 10]. This has been largely due to the absence of an Env trimer structure, which has been thwarted by its lack of stability as a soluble protein and extensive N-linked glycosylation (∼ 81 sites per trimer). During Env synthesis, the gp160 precursors trimerize and are subsequently cleaved by proteases of the furin

http://www.worldscientific.com/worldscibooks/10.1142/ family into gp120 and gp41 subunits, which associate non-covalently. However, attempts to isolate Env trimers have generally failed due to aggregation, instability and lack of proper folding of uncleaved trimers [11–14]. The current view of Env- mediated entry involves a two-step process that is initiated by the binding of gp120 to the CD4 receptor on T cells. This encounter induces conformational changes in Env that result in interactions with a co-receptor (generally CXCR4 or CCR5) that triggers further conformational changes in the gp41 transmembrane segment

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World that leads to formation of an extended coil-coil pre-fusion intermediate [15]. Sub- sequently, the resulting N-terminal peptide of gp41 is inserted and anchored into NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY host cell membrane. The gp41 then forms a six-helix bundle that brings the viral and host membranes together.

Thus, understanding the function of viral pathogens has been facilitated by structure determination of their surface antigens but it has been a challenge for HIV-1, HCV and until recently for RSV [16] and Ebola [17] for example. However, use of diﬀerent expression systems, manipulation of the glycosylation, stabilization October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 295

HIV Envelope and Inﬂuenza Hemagglutinin Fusion Glycoproteins 295

of the antigens, and forming complexes with antibodies has provided opportunities for high resolution structural studies. However, capturing fusion intermediates has not been possible or visualizing the molecular events that lead to fusion between

the viral membrane and cell membrane.

My recent research contributions to structure and function of viral glycoproteins, the structural bass of neutralization, and universal vaccines My laboratory focuses on understanding the interaction and neutralization of mi- distribution is allowed. distribution

crobial pathogens by the immune system through high-resolution X-ray structural

studies of antibodies complexes with viral antigens. Recently, we have focused on structural investigations of the key sites of vulnerability on inﬂuenza and HIV-1 9180 viruses with the ultimate goal of developing sustainable cross-serotype immune re- only. No further No only.

sponses [18]. Antibody-mediated neutralization of inﬂuenza and HIV-1 viruses is a complex combinatorial problem for the human immune system as it is presented with diverse, highly variable and constantly evolving viruses. While neutralizing antibodies against these viruses are traditionally regarded as being strain speciﬁc,

vayinstitutes.be/ studies have shown that a much broader response can be mounted. While these PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

examples provide compelling evidence that the immune system is capable of mount- - ing a sustained, cross-serotype response against these viruses, how to elicit broadly neutralizing antibodies by vaccination is poorly understood. In that optic, we are

http://www.sol investigating the structures of the viral antigens and the basis of broad neutralization by antibodies. We have delineated sites of vulnerability on the HA and Env

host on on host to enable development of novel vaccine scaﬀolds and even small molecule inhibitors that may ameliorate or prevent disease progression [7, 19–21]. To address this challenging task, we have used an integrative approach using combined biophysical and biochemical methodologies in close collaboration with the other laboratories

http://www.worldscientific.com/worldscibooks/10.1142/ at Scripps and elsewhere by investigating complexes of viral glycoproteins with broadly neutralizing antibodies. We employ state-of-the-art X-ray crystallography, electron microscopy and glycan array technologies to provide key insights into in- ﬂuenza virus and HIV-1 neutralization, tropism and pathogenesis, to reveal novel strategies to control and combat future pandemics. Most importantly, we have high quality reagents for soluble hemagglutinin [8] and HIV-1 Env trimer [11, 12, 14] and gp120 etc.

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Structural characterization of both HIV-1 Env and inﬂuenza HA in complex with an extensive toolbox of diverse antibodies has been very informative at deﬁning NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY the key sites of vulnerability on these viruses [18, 22–24]. These analyses have clearly succeeded in uncovering novel modes and sites of antibody recognition, not

anticipated from previous studies. The epitopes on HIV-1 are more diverse in type and number and some of the most potent sites of neutralization depend on both glycans and protein [18, 25–31]. We are exploring whether identiﬁcation of these sites of vulnerability can be directly utilized in vaccine settings. Interestingly, bnAbs October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 296

296 I. A. Wilson

have been identiﬁed to both ﬂu and HIV-1 that bind to key functional regions of each viral spike: the receptor binding site and the stem region proximal to the viral membrane involved in the fusion machinery. Antibody attachment to the

receptor binding site most likely leads to potent neutralization by directly preventing

receptor engagement on host cells, whereas binding to a region proximal to the viral membrane appears to prevent membrane fusion. The information derived from targeting these sites by diﬀerent antibodies can now be used also for small molecule [18, 23, 32, 33] and small protein inhibitors [34].

distribution is allowed. distribution Outlook for future developments on HIV and Inﬂuenza vaccines

While the challenges presented by HIV-1, inﬂuenza and other incredibly evasive pathogens are daunting, the presence of high titers of neutralizing antibodies, some 9180 which are broadly neutralizing, in subsets of infected individuals suggest that the only. No further No only.

repertoire of the human immune system are up to the challenge. Whether modern vaccine design can appropriately elicit and harness this potential remains to be seen. The current structural arsenal of antibodies that cover both inﬂuenza A and inﬂuenza B and the availability of multiple structures of antibodies in complex with

vayinstitutes.be/ HA that target the same functionally constrained epitopes in the receptor binding PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- site and stem, indicate that design of a universal broad-spectrum vaccine in the near future seems very promising. For HIV-1, we are working on a better structural understanding of the HIV-1 Env trimer to guide the design of more appropriate

http://www.sol immunogens that possess all of the characteristics and constraints necessary to present the epitopes of broadly neutralizing antibodies in a more native context.

host on on host Scaffolded epitopes are also being designed for use as immunogens to elicit specific and focused responses against specific epitopes [25, 35–39]. The challenge still remains are to how to use these immunogens to stimulate potent and broad B cell responses that mimic natural infection and, in the case of HIV-1, to elicit a faster

http://www.worldscientific.com/worldscibooks/10.1142/ response than the two years it takes to generate a broad neutralizing response in humans after HIV-1 infection [40–42].

Acknowledgments Marc Elsliger and Jean-Philippe Julien are gratefully acknowledged for help with preparation of this manuscript as well as the NIH, the International AIDS Vaccine

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Initiative Neutralizing Antibody Consortium and Center, and the CHAVI-ID for support. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

References

1. I. A. Wilson, J. J. Skehel, D. C. Wiley, Nature 289, 366 (1981). 2. J. J. Skehel, D. C. Wiley, Annu. Rev. Biochem. 69, 531 (2000). 3. Y. Okuno, Y. Isegawa, F. Sasao, S. Ueda, J. Virol. 67, 2552 (1993). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 297

HIV Envelope and Inﬂuenza Hemagglutinin Fusion Glycoproteins 297

4. M. Throsby, E. van den Brink, M. Jongeneelen, L. L. Poon, P. Alard et al., PLoS One 3, e3942 (2008). 5. A. K. Kashyap, J. Steel, A. F. Oner, M. A. Dillon, R. E. Swale et al., Proc.

Natl. Acad. Sci. USA 105, 5986 (2008).

6.J.Sui,W.C.Hwang,S.Perez,G.Wei,D.Airdet al., Nat. Struct. Mol. Biol. 16, 265 (2009). 7. D. C. Ekiert, I. A. Wilson, Curr. Opin. Virol. 2, 134 (2012). 8.D.C.Ekiert,G.Bhabha,M.A.Elsliger,R.H.Friesen,M.Jongeneelenet al., Science 324, 246 (2009). 9. S. A. Gallo, C. M. Finnegan, M. Viard, Y. Raviv, A. Dimitrov et al., Biochim. distribution is allowed. distribution Biophys. Acta 1614, 36 (2003).

10.I.Markovic,K.A.Clouse,Curr. HIV Res. 2, 223 (2004). 11. R. Khayat, J. H. Lee, J. P. Julien, A. Cupo, P. J. Klasse et al., J. Virol. 87, 9180 9865 (2013). only. No further No only. 12. P. J. Klasse, R. S. Depetris, R. Pejchal, J. P. Julien, R. Khayat et al., J. Virol. 87, 9873 (2013). 13.J.M.Binley,R.W.Sanders,B.Clas,N.Schuelke,A.Masteret al., J. Virol. 74, 627 (2000).

vayinstitutes.be/ 14.R.Ringe,R.W.Sanders,A.Yasmeen,H.J.Kim,J.H.Leeet al., Proc. Natl. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Acad. Sci. USA 110, 18256 (2013). 15. I. Mu˜noz-Barroso, K. Salzwedel, E. Hunter, R. Blumenthal, J. Virol. 73, 6089 (1999). http://www.sol 16. J. S. McLellan, M. Chen, S. Leung, K. W. Graepel, X. Du et al., Science 340, 1113 (2013). host on on host 17. J. E. Lee, M. L. Fusco, A. J. Hessell, W. B. Oswald, D. R. Burton et al., Nature 454, 177 (2008). 18. J. P. Julien, P. S. Lee, I. A. Wilson, Immunol. Rev. 250, 180 (2012). 19. D. R. Burton, R. Ahmed, D. H. Barouch, S. T. Butera, S. Crotty et al., Cell

http://www.worldscientific.com/worldscibooks/10.1142/ Host Microbe 12, 396 (2012). 20. D. R. Burton, P. Poignard, R. L. Stanﬁeld, I. A. Wilson et al., Science 337, 183 (2012). 21.M.J.vanGils,R.W.Sanders,Virology 435, 46 (2013). 22. P. D. Kwong, I. A. Wilson, Nat. Immunol. 10, 573 (2009). 23.M.Hong,P.S.Lee,R.M.Hoﬀman,X.Zhu,J.C.Krauseet al., J. Virol. 87, 12471 (2013).

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 24. N. S. Laursen, I. A. Wilson, Antiviral. Res. 98, 476 (2013).

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 25. J. Jardine, J. P. Julien, S. Menis, T. Ota, O. Kalyuzhniy et al., Science 340, 711 (2013). 26.J.P.Julien,J.H.Lee,A.Cupo,C.D.Murin,R.Derkinget al., Proc. Natl. Acad. Sci. USA 110, 4351 (2013). 27. J. P. Julien, D. Sok, R. Khayat, J. H. Lee, K. J. Doores et al., PLoS Pathog. 9, e1003342 (2013). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 298

298 I. A. Wilson

28. L. Kong, J. H. Lee, K. J. Doores, C. D. Murin, J. P. Julien et al., Nat. Struct. Mol. Biol. 20, 796 (2013). 29. J. S. McLellan, M. Pancera, C. Carrico, J. Gorman, J. P. Julien et al., Nature

480, 336 (2011).

30. R. Pejchal, K. J. Doores, L. M. Walker, R. Khayat, P. S. Huang et al., Science 334, 1097 (2011). 31. L. J. Ross, M. M. Binns, J. Pastorek, J. Gen. Virol. 72, 949 (1991). 32. R. Xu, J. C. Krause, R. McBride, J. C. Paulson, J. E. Crowe Jr. et al., Nat. Struct. Mol. Biol. 20, 363 (2013). 33. P. S. Lee, R. Yoshida, D. C. Ekiert, N. Sakai, Y. Suzuki et al., Proc. Natl. Acad. distribution is allowed. distribution Sci. USA 109, 17040 (2012).

34. S. J. Fleishman, T. A. Whitehead, D. C. Ekiert, C. Dreyfus, J. E. Corn et al., Science 332, 816 (2011). 9180 35. M. L. Azoitei, B. E. Correia, Y. E. Ban, C. Carrico, O. Kalyuzhniy et al., only. No further No only. Science 334, 373 (2011). 36. B. E. Correia, Y. E. Ban, M. A. Holmes, H. Xu, K. Ellingson et al., Structure 18, 1116 (2010). 37. J. Guenaga, P. Dosenovic, G. Ofek, D. Baker, W. R. Schief et al., PLoS One

vayinstitutes.be/ 6, e16074 (2011). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 38. G. Ofek, F. J. Guenaga, W. R. Schief, J. Skinner, D. Baker et al., Proc. Natl. Acad. Sci. USA 107, 17880 (2010). 39. R. L. Stanﬁeld, J. P. Julien, R. Pejchal, J. S. Gach, M. B. Zwick et al., J. Mol. http://www.sol Biol. 414, 460 (2011). 40. N. A. Doria-Rose, R. M. Klein, M. M. Manion, S. O’Dell, A. Phogat et al., J. host on on host Virol. 83, 188 (2009). 41. D. N. Sather, J. Armann, L. K. Ching, A. Mavrantoni, G. Sellhorn et al., J. Virol. 83, 757 (2009). 42. M. D. Simek, W. Rida, F. H. Priddy, P. Pung, E. Carrow et al., J. Virol. 83,

http://www.worldscientific.com/worldscibooks/10.1142/ 7337 (2009). ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 299

299

DECONSTRUCTION OF ITERATIVE POLYKETIDE SYNTHASES

CRAIG A. TOWNSEND Department of Chemistry, The Johns Hopkins University, 3400 N. Charles Street Baltimore, MD 21218-2685, USA

The enormous progress of the last 25 years in natural products biochemistry has distribution is allowed. distribution been dominated by investigation of giant microbial enzymes comprised of orga-

nized catalytic domains that function according to two distinct paradigms. Each

9180 is “programmed” to sequentially assemble simple precursor molecules over multiple steps without diﬀusible intermediates to synthesize highly elaborated products. only. No further No only. Both synthetic strategies rely upon carrier proteins to hold and shuttle extending intermediates among catalytic domains in an ordered fashion. Collectively these megaproteins are impressive “supramolecular machines” exemplifying protein catalysis far more sophisticated than the classical view of “one gene, one enzyme, one vayinstitutes.be/ reaction” [1], and are of central importance to the creation of small-molecule drugs PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- for human and animal health.

Assembly-line and iterative catalysis in natural product http://www.sol biosynthesis

host on on host The application of molecular biological techniques to natural product biosynthetic problems in the mid 1980’s ushered in the modern era of research in the ﬁeld. Seminal events include isolation of the penicillin gene cluster and characterization of the non-ribosomal peptide synthetase (NRPS) ACVS that assembles the classi-

http://www.worldscientific.com/worldscibooks/10.1142/ cal Arnstein L,L,D-tripeptide precursor of the antibiotic [2], Fig. 1. Biochemical experiments on isolated wild-type NRPS enzymes had already established a linear mode of accreting ATP-dependent synthesis [3], but now isolation of the corresponding biosynthetic genes and their translation to domains of recognized function in loosely repeated sets, or “modules,” underscored the exquisitely organized nature of the “assembly line” synthetic program executed by these large enzymes. The end of 1990 marked the watershed co-discovery of genes encoding three giant modular

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World polyketide synthases (PKSs) that synthesize the 6-deoxyerythronolide core of the macrolide antibiotic erythromycin [4, 5]. Together deoxyerythronolide B synthases NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY (DEBS)1-3, totaling ca. 106 Da, opened the door to a second paradigm of assembly line megaenzymes whose synthetic program could be “read” from the N-to

C-terminus. The parallel universes of NRPS and modular PKS enzymes in their canonical forms (there are many exceptions and variations on the theme) are extravagant ex- tensions of the classical view of enzyme function enunciated 70 years ago [1]; that October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 300

300 C. A. Townsend

is, each domain of the assembly line carries out a single chemical reaction and the carrier protein delivers the extending intermediates to the next programmed domain and so on to the ﬁnal product release step. Among the superfamily of PKSs,

however, the modular, assembly line enzymes are but one subtype. As shown in

Table 1, four other families of Type I, or multidomain proteins, are known emblem- atic of the second programmatic strategy of “iterative” catalysis. Iterative catalysis fundamentally diﬀers from the assembly line paradigm where a smaller set, often the minimal set, of catalytic domains is present, but their active sites are reused a ﬁxed, or “programmed”, number of times. distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- http://www.sol host on on host Fig. 1. Representative “assembly line” natural products. The non-ribosomal peptide synthetase (NRPS) Arnstein tripeptide from ACV synthetase and that from the modular polyketide synthases (PKSs) deoxyerythronolide B synthases (DEBS) 1-3. http://www.worldscientific.com/worldscibooks/10.1142/

Iterative polyketide synthases as supramolecular machines Fatty acid synthases (FASs) are eﬀectively one subgroup of the Type I, iterative megaenzymes in Table 1 that share with the PKSs the use of acylCoA precursors and identical mechanisms of two-carbon chain extension, but diﬀer in the extent of β- carbon processing. At one extreme are the FASs where reductions and dehydration

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World occur at every β-carbonyl to produce a fully saturated fatty acid product. At the other lie the fully non-reducing polyketide synthases (NR-PKSs) where the same NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY chain elongation machinery exists, but no further modiﬁcations occur at the β- carbon to generate a poly-β-ketone intermediate of “programmed” length, which

is cyclized in specialized domains to aromatic products, Fig. 2 [6]. The aromatic and polycyclic aromatic products, pigments and toxins of fungi are some of the oldest natural products known. Ironically the function of modular PKSs, despite the greater structural complexity of their products, could be readily established October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 301

Deconstruction of Iterative Polyketide Synthases 301

by either inactivation of β-processing in speciﬁc modules or by inserting the chain- terminating thioesterase domain at the ends of internal modules to yield predictably truncated products. Iterative PKSs, however, remained shrouded in mystery and

technical barriers. Inactivation of a synthetic domain in an iterative system stops

all synthesis and next to nothing can be learned. A new approach was needed. A series of contributions from my laboratory over the last dozen years has opened iterative PKSs to systematic investigation using methods that can be more widely applied. Our focus has been on the NR-PKS PksA, which is central to the biosynthesis of the potent environmental toxin and carcinogen aﬂatoxin B1. First, we developed a bioinformatics algorithm, UMA, that can identify linker regions in a distribution is allowed. distribution group of related polydomain proteins [7]. Application of UMA to PksA yielded the

surprising result that unlike the four catalytic domains readily identiﬁed by conventional sequence analysis, a β-ketoacyl synthase (KS) domain, a malonyl acyltrans- 9180 ferase (MAT) domain, and acyl-carrier protein (ACP) and a C-terminal thioesterase only. No further No only. (TE) domain, a large, well deﬁned N-terminal domain was clearly visible, and an unknown internal domain placed between the MAT and ACP domains. Were these functional in some way, or merely structural? Using UMA as a guide, PksA at the gene level was cut (dissection) to yield

vayinstitutes.be/ mono-, di- and tridomains as soluble, functional proteins in E. coli that could be PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- recombined (reconstitution) to deduce, or “deconstruct” [8], the synthetic role of each domain. The N-terminal domain we discovered is a “starter-unit acyltrans- ferase” (SAT) domain [9], which in most NR-PKSs accepts acetylCoA from cellular http://www.sol primary metabolism and loads the PKS for synthesis [10]. In the case of PksA, however, a more specialized situation obtains that PksA has a closely associated host on on host yeast-like pair of fatty acid synthase subunits that prepare a hexanoyl starter unit 6 that is transferred directly to the PksA SAT domain in a ∼1.3 x 10 Da α2β2γ2 complex to prime polyketide extension (Fig. 2) [11, 12]. Like the SAT domain, which bore no compelling sequence similarity to any other

http://www.worldscientific.com/worldscibooks/10.1142/ known protein (except other similarly placed domains in established and suspected NR-PKSs), the well-deﬁned internal domain also showed no useful sequence homology. Deconstruction experiments, however, decisively demonstrated that it binds the fully elongated C20-poly-β-ketone and controls the pluripotent self-condensation of this reactive intermediate to a single bicyclic product and, thus, was dubbed a “product template” (PT) domain. This bicyclic intermediate was then transferred from the ACP to the TE domain, not for typical hydrolytic release, but for a third,

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Dieckmann, cyclization to simultaneously release the product norsolorinic acid an-

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY throne (Fig. 2), the ﬁrst dedicated intermediate en route to aﬂatoxin [8]. X-ray crystal structures of the PksA PT and TE domains, the product determining cyclization catalysts, and molecular modeling have led to hypotheses to account for how these key cyclization steps occur [13, 14]. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 302

302 C. A. Townsend

distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Fig. 2. Comparison of animal fatty acid synthases and non-reducing polyketide synthases, which lack any β-processing domains but contain unique starter unit transacylase (SAT) and product http://www.sol template (PT) domains originally discovered by UMA analysis of PksA. The highly complex aﬂatoxin biosynthetic pathway is initiated by a complex of yeast-like fatty acid synthase subunits

host on on host HexA and HexB, which form a complex with PksA and carry out the synthesis of norsolorinic acid anthrone, the ﬁrst committed intermediate to the mycotoxin.

Outlook •

http://www.worldscientific.com/worldscibooks/10.1142/ Fundamental issues of “programming” control in iterative systems of chain length, control of β-carbon processing during polyketide elongation, and control of cyclization events are in most respects unknown. • The quarternary structures of these large, believed dimeric proteins remain important unsolved problems, although fragmentary information exists for mono- and didomains. As Professor Ban will address, the work of his and the Steitz groups on animal and yeast fatty acid synthases are pioneering successes and conceptual

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World models for the ﬁeld. • Fragmentary information exists that domain • domain interactions aﬀect “pro- NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY gramming”. For example, recent observations in PksA suggest that the KS domain “waits” for the correct starter, a hexanoyl group as opposed to acetyl, to be bound at

the active site cysteine before polyketide extension is initiated, despite the demon- strable presence of acetyl groups on the ACP during synthesis [8, 15]. • The orchestration of 20, 30, 50 or more reactions without release of intermediates, but the total synthesis of a discrete product is an impressive hallmark of October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 303

Deconstruction of Iterative Polyketide Synthases 303

iterative catalysis where active sites are reused multiple times and each, therefore, must accommodate a series of structurally distinct intermediates as the polyketide chain extends. The “editing” behavior of TE domains is key to faithful and eﬃcient

synthesis [15, 16]. So too are balanced internal kinetics of synthesis and transfer

among active sites to maintain synthesis in the forward direction [15]. The roles of protein dynamics, inter-domain cooperative eﬀects (positive and negative), and the possibility of “triggering” initiated by substrate (the correct substrate) binding to advance correctly edited intermediates onward to product all present daunting problems for study. • Necessary formation of higher order catalytic complexes is increasingly recognized distribution is allowed. distribution in natural product biosynthesis where enzymes require “auxiliary” or “helper” pro-

teins to function eﬃciently, if at all. The recently elucidated role of the MbtH superfamily activation of some, but not all, adenylation domains in NRPSs is a 9180 shining example of their importance [17]. In trans interactions of this sort will be only. No further No only. found to play newly appreciated roles in the activities of NRPS and PKS enzymes. • Finally, engineering and directed evolution of biosynthetic enzymes is in its infancy. Iterative systems are especially attractive because of the already promiscuous nature for all participating synthetic domains. Unlike modular systems where active

vayinstitutes.be/ sites typically function once during chain elongation, iterative catalytic domains are PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- used multiple times to faithfully shepherd substrates having a remarkable breadth of molecular sizes to defined products. Possibilities can be envisioned where their intrinsic substrate flexibility can be exploited. What is lost in binding affinity may http://www.sol well be compensated for by high effective concentration provided by carrier-protein mediated intraprotein transfer [18]. host on on host

Acknowledgments The National Institutes of Health are gratefully acknowledged for their sustained support. http://www.worldscientific.com/worldscibooks/10.1142/

References 1. G. W. Beadle, E. L. Tatum, Proc. Natl. Acad. Sci. USA 27, 499 (1941). 2. N. M. Sampson, R. Belagaje, D. T. Blankenship, J. L. Chapman, D. Perry et al., Nature 318, 191 (1985). 3. F. Lipmann, Acc. Chem. Res. 6, 361 (1973).

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 4.J.Cortes,S.F.Haydock,G.A.Roberts,D.J.Bevitt,P.F.Leadlay,Nature 348, 176 (1990). NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 5. S. Donadio, M. J. Staver, J. B. McAlpine, S. J. Swanson, L. Katz, Science 252, 675 (1991).

6. J. M. Crawford, C. A. Townsend, Nat. Rev. Microbiol. 8, 879 (2010). 7. D. W. Udwary, M. Merski, C. A. Townsend, J. Mol. Biol. 323, 585 (2002). 8. J. M. Crawford, P. M. Thomas, J. R. Scheerer, A. L. Vagstad, N. L. Kelleher et al., Science 320, 243 (2008). October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 304

304 C. A. Townsend

9. J. M. Crawford, B. C. R. Dancy, E. A. Hill, D. W. Udwary, C. A. Townsend, Proc. Natl. Acad. Sci. USA 103, 16728 (2006). 10. J. M. Crawford, A. L. Vagstad, K. R. Whitworth, K. C. Ehrlich, C. A.

Townsend, ChemBioChem 9, 1019 (2008).

11. C. M. H. Watanabe, C. A. Townsend, Chem. Biol. 9, 981 (2002). 12. J. Foulke-Abel, C. A. Townsend, ChemBioChem 13, 1880 (2012). 13. J. M. Crawford, T. P. Korman, J. W. Labonte, A. L. Vagstad, E. A. Hill et al., Nature 461, 1139 (2009). 14. T. P. Korman, J. M. Crawford, J. W. Labonte, A. G. Newman, J. Wong et al., Proc. Natl. Acad. Sci. USA 107, 6246 (2010). distribution is allowed. distribution 15. A. L. Vagstad, S. B. Bumpus, K. Belecki, N. L. Kelleher, C. A. Townsend, J.

Amer. Chem. Soc. 134, 6865 (2012). 16. A. G. Newman, A. L. Vagstad, K. Belecki, J. R. Scheerer, C. A. Townsend, 9180 Chem. Commun. 48, 11772 (2012). only. No further No only. 17. E. A. Felnagle, J. J. Barkei, H. Park, A. M. Podevels, M. D. McMahon et al., Biochemistry 49, 8815 (2010). 18. A. L. Vagstad, A. G. Newman, P. A. Storm, K. Belecki, J. M. Crawford et al., Angew. Chem. Int. Ed. 51, 1 (2013). vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 305

305

REGULATING RIBOSOME PAUSING DURING TRANSLATION

MARINA V. RODNINA Department of Physical Biochemistry, Max Planck Institute for Biophysical Chemistry Am Fassberg 11, Goettingen, 37077, Germany distribution is allowed. distribution

Ribosome processivity and translational pauses

9180 In bacteria, the average rate of translation elongation is 10–20 amino acids per second and the error frequency ranges between 10−4-10−3 [1, 2]. The overall rate only. No further No only. of translation elongation can be inﬂuenced by many factors, such as the rate of the codon-speciﬁc delivery of cognate aminoacyl-tRNAs (aa-tRNAs) to the A site of the ribosome, the abundance of the respective tRNAs, the chemistry of peptide bond formation, secondary structure elements in the mRNA, mRNA context, controlled

vayinstitutes.be/ ribosome stalling, collisions between individual ribosomes in polysomes, and the PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- cooperation between translating ribosomes and the RNA polymerase machinery. As a result of this complex interplay, some codons are translated faster or slower than the average, i.e., the rate of translation is not uniform along an mRNA. Due to the http://www.sol limited information, the rules that define the local speed and accuracy of translation in different regions of a given mRNA are not fully understood [3]. Usually, speed host on on host and accuracy of protein synthesis are tightly controlled throughout the translation process; however, the consequences of changes in translational efficiency on the cellular level, and the long-term evolutionary responses to such changes, are not known [4]. For a given amino acid sequence of a protein, multiple degrees of freedom

http://www.worldscientific.com/worldscibooks/10.1142/ still remain that may allow evolution to tune efficiency and fidelity of gene expression under various conditions [3]. Thus, the sequences on the mRNA and the tRNA pool provide an additional, rich informational framework that may define the proteome of the cell beyond the rules of the genetic code.

Stalling at proline sequences is alleviated by EF-P

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Kinetic measurements suggested that peptide bond formation with P-site peptidyl- tRNAs ending with proline or with Pro-tRNAPro in the decoding site of the ri- NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY bosome turned out to be surprisingly slow [5, 6], which might lead to stalling of translation as a whole. In fact, translating ribosomes tend to be stalled on mRNA

sequences coding for two consecutive prolines followed by Asn, Asp, Glu, Gly, Pro, or Trp [7]. We discovered that unlike peptide bond formation with the other amino acids, translation of those sequences is accelerated by a specialized translation factor, elongation factor P (EF-P) [5, 8]. Due to the presence of the pyrrolidine ring October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 306

306 M. V. Rodnina

in proline, the torsion angle for the N–Cα bond restricts the number of accessible conformations and imposes structural constraints on Pro positioning in the peptidyl transferase center of the ribosome, thereby sterically hindering peptide bond

formation. Of more than 4000 annotated proteins in Escherichia coli, > 300 con-

tain motifs of three or more consecutive prolines or X-Pro-Pro-X motifs, where X is one of the stalling-promoting amino acids [9]. Among those proteins metabolic enzymes, transporters, and regulatory transcription factors are frequent, explaining the pleiotropic phenotypes caused by deletions of the genes coding for EF-P or its modification enzymes, including effects on virulence and bacterial fitness [10]. Given that the synthesis of a peptide bond usually does not require an auxiliary factor, distribution is allowed. distribution EF-P appears to be a recent evolutionary addition to the repertoire of translation

factors, possibly required to properly balance the cellular amounts of regulatory proteins containing proline repeats. 9180 EF-P carries a unique post-translational modification which requires the action only. No further No only. of three enzymes, YjeK, YjeA, and YcfM [10–13]. The low catalytic proficiency of unmodified EF-P may explain why deletions of the yjeA or yjeK genes that code for the modification enzymes lead to phenotypes that are similar to, or only somewhat milder than, the deletion of efp, the gene coding for EF-P [14, 15]. On

vayinstitutes.be/ the other hand, even in the absence of active EF-P, small amounts of proteins with PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- proline stretches are slowly formed; this may be sufficient to support the viability of strains in which the efp, yieA,andyieK genes are deleted. The modification enzymes, which are lacking in higher eukaryotes, provide attractive new targets for http://www.sol the development of new, highly specific antimicrobials. host on on host

EF-P and evolution Of the 21 proteinogenic amino acids, Pro (together with Ala, Glu, and Gly) tends to be lost during the evolution [16], possibly because it stalls the translation. The http://www.worldscientific.com/worldscibooks/10.1142/ emergence of a factor that specifically accelerated Pro incorporation might result from an evolutionary trade-off between the crucial role of Pro clusters in some proteins and the requirement for rapid, processive protein synthesis. One can en- visage that the specific requirements of eukaryotic cells, particularly with respect to translational control, compartmentalization of protein synthesis, and larger protein domains, might lead to emergence of additional, yet uncharacterized proteins that modify the functions of the ribosome or of the universal translation factors ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World [17]. Identification and understanding of this auxiliary machinery would be a very NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY exciting avenue for future research.

Acknowledgments The work was funded by the Max Planck Society and by a grant of the Deutsche Forschungsgemeinschaft. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 307

Regulating Ribosome Pausing During Translation 307

References 1. M. V. Rodnina, Ribosomes: Structure, Function, and Dynamics,eds.W.Win- termeyer, R. Green (Springer, Wien, New York), 199 (2011).

2. I. Wohlgemuth, C. Pohl, J. Mittelstaet, A. L. Konevega, M. V. Rodnina, Philos. Trans. R. Soc. Lond. B Biol. Sci. 366, 2979 (2011). 3. H. Gingold, Y. Pilpel, Mol. Syst. Biol. 7, 481 (2011). 4. D. A. Drummond, C. O. Wilke, Nat. Rev. Genet. 10, 715 (2009). 5. L. K. Doerfel, I. Wohlgemuth, C. Kothe, F. Peske, H. Urlaub et al., Science 339, 85 (2013). 6. I. Wohlgemuth, S. Brenner, M. Beringer, M. V. Rodnina, J. Biol. Chem. 283, distribution is allowed. distribution 32229 (2008). 7. C. J. Woolstenhulme, S. Parajuli, D. W. Healey, D. P. Valverde, E. N. Petersen

9180 et al., Proc. Natl. Acad. Sci. USA 110, E878 (2013). 8. S. Ude, J. Lassak, A. L. Starosta, T. Kraxenberger, D. N. Wilson et al., Science only. No further No only. 339, 82 (2013). 9. L. Peil, A. L. Starosta, J. Lassak, G. C. Atkinson, K. Virumae et al., Proc. Natl. Acad. Sci. USA 110, 15265 (2013). 10.W.W.Navarre,S.B.Zou,H.Roy,J.L.Xie,A.Savchenkoet al., Mol. Cell vayinstitutes.be/ 39, 209 (2010). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 11. L. Peil, A. L. Starosta, K. Virumae, G. C. Atkinson, T. Tenson et al., Nat. Chem. Biol. 8, 695 (2012). 12. H. Roy, S. B. Zou, T. J. Bullwinkle, B. S. Wolfe, M. S. Gilreath et al., Nat. http://www.sol Chem. Biol. 7, 667 (2011). 13. T. Yanagisawa, T. Sumida, R. Ishii, C. Takemoto, S. Yokoyama, Nat. Struct. host on on host Mol. Biol. 17, 1136 (2010). 14.S.B.Zou,S.J.Hersch,H.Roy,J.B.Wiggers,A.S.Leunget al., J. Bacteriol. 194, 413 (2012). 15. S. B. Zou, H. Roy, M. Ibba, W. W. Navarre, Virulence 2, 147 (2011). http://www.worldscientific.com/worldscibooks/10.1142/ 16. I. K. Jordan, F. A. Kondrashov, I. A. Adzhubei, Y. I. Wolf, E. V. Koonin et al., Nature 433, 633 (2005). 17. L. K. Doerfel, M. V. Rodnina, Biopolymers 99, 837 (2013). ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 308

308

THE MOLECULAR MECHANICS OF THE RIBOSOME

JIE ZHOU, LAURA LANCASTER, ZHUOJUN GUO, JOHN PAUL DONOHUE and HARRY F. NOLLER Department of Molecular, Cell and Developmental Biology and Center for Molecular Biology of RNA, University of California at Santa Cruz, Santa Cruz, CA 95064, USA distribution is allowed. distribution

Research on the ribosome as a supramolecular machine 9180 Ribosomes, which are responsible for synthesis of proteins in all organisms, are only. No further No only.

among the largest and most complex of biological supramolecular machines, comprising several dozens of protein and RNA molecules. During the past two decades, our understanding of the structure of the ribosome has made spectacular progress, ﬁrst from cryo-EM and then from X-ray crystallography. With all-atom structures

vayinstitutes.be/ for examples of both bacterial and eukaryotic ribosomes in hand, the stage is set PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

for understanding the mechanisms of protein synthesis at the atomic level. Funda- - mental questions that have gone unanswered for several decades are now beginning to be resolved. Protein synthesis involves translation of the nucleotide sequences of

http://www.sol messenger RNAs into proteins via the genetic code. Although the detailed mechanisms of initiation and termination of translation diﬀer between bacterial, archaeal

host on on host and eukaryal systems, the mechanism of elongation is similar for all organisms. Each round of the elongation cycle takes place in three steps: (i) selection of aminoacyl-tRNA, (ii) catalysis of peptide bond formation and (iii) coupled translocation of mRNA and tRNA. Each of these steps is believed to be based mainly

http://www.worldscientific.com/worldscibooks/10.1142/ on the properties of ribosomal RNA (rRNA), rather than the ribosomal proteins, in keeping with the notion that the ﬁrst ribosomes had to be made only of RNA. Aminoacyl-tRNA selection, based on codon-anticodon base pairing, is monitored by interactions with universally conserved bases of 16S rRNA in the decoding site of the small ribosomal subunit [1]. Peptide bond formation is catalyzed by peptidyl transferase, an enzymatic activity of 23S rRNA in the large ribosomal subunit [2]. Although translocation of mRNA and tRNA is catalyzed by the protein elonga-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World tion factor EF-G, it seems likely that its fundamental mechanism is again based on ribosomal RNA. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Following peptide bond formation, the mRNA must be moved through the ribosome by one codon. In addition, the deacylated P-site tRNA and peptidyl A-site

tRNA must be moved from the A and P sites to the P and E sites, respectively. These represent large-scale movements of 20 Ato70˚ A,˚ which take place on a time scale of about 50 ms. Chemical probing experiments showed that tRNA movement occurs in two steps, involving hybrid binding states [3]. In the ﬁrst step, the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 309

The Molecular Mechanics of the Ribosome 309

A-site peptidyl-tRNA moves into the A/P hybrid state and the deacylated P-site tRNA moves into the P/E hybrid state. In the second step, the A/P tRNA moves into the P/P state (the classical P site) and the P/E tRNA moves into the E/E

state (the classical E site). The ﬁrst step is reversible and can occur spontaneously,

while the second, rate-limiting step depends on catalysis by EF-G.

Studying ribosome mechanics by crystallography and FRET Cryo-EM studies by Frank and Agrawal showed that when EF-G binds to the ribosome, the small ribosomal subunit rotates by 5–10◦ counterclockwise relative to distribution is allowed. distribution the large subunit [4]. In a more recent cryo-EM study from the Spahn laboratory,

a ribosome complex trapped with fusidic acid in an intermediate state of translocation, contained a tRNA bound in a position in between the P/E and E/E states 9180 [5]. In this complex, the head of the 30S subunit was rotated by about 18◦, suggest- only. No further No only.

ing that the rate-limiting step of translocation, in which the mRNA and anticodon stem-loop moieties are moved through the small subunit, is accomplished by rotation of the 30S head. We sought to (a) obtain direct evidence for rotation of the 30S head during translocation and (b) to solve the crystal structure of a trapped

vayinstitutes.be/ translocation intermediate. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

We used ensemble FRET methods to study 30S head rotation in real time [7]. - We labeled speciﬁc positions of ribosomal proteins in the head and body of the 30S subunit by in vitro reconstitution with ﬂuorophore-labeled proteins. The in- http://www.sol

(a) (b) host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Fig. 1. Rotation of the 30S subunit head during EF-G-catalyzed translocation [6]. (a) mRNA translocation, as followed by quenching of ﬂuorescence of a pyrene dye attached to position +9 of mRNA. (b) Rotation of 30S subunit head, followed by changes in FRET eﬃciency between donor

Alexa 488 and acceptor Alexa 568 probes attached to position 101 of S11 and position 11 of S13 for the S11-D/S13-A pair (green), and position 108 of S12 and position 56 of S19 for the S12-D/S19-A pair (red). Pre-translocation 70S ribosome complexes were rapidly mixed with EF-G-GTP at time = 0 and pyrene ﬂuorescence (a) or Alexa 568 emission (b) were recorded. Each trace represents the average of ten reactions. WT: wild-type, unlabeled ribosomes; AU: arbitrary units. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 310

310 J. Zhou et al.

termolecular distance between proteins S19 (head) and S12 (body) was predicted to increase, and the distance between S13 (head) and S11 (body) was predicted to decrease during translocation. These predictions were conﬁrmed (Fig. 1) using a

rapid mixing stopped-ﬂow approach. The forward rate of head rotation was slower

than the forward rate of 30S subunit body rotation [8], and the reverse rate of head rotation was roughly similar to the reverse rate of 30S body rotation. Our results are consistent with coupling of 30S head rotation to the second step of translocation.

(a) (b) distribution is allowed. distribution

9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- (c) (d) (e) http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/ (f) (g) (h)

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Fig. 2. Structures of trapped 70S ribosome·EF-G complexes [6]. (a,b) Overall views of the (a) non-rotated 70S·EF-G-post complex [10] and (b) Fus 70S·EF-G complex with tRNA bound in the pe∗/E state. (c–e) Interface views showing 30S subunit body and head rotation in the

(c) EF-G-post state; (d) GDPNP-I and Fus complex; and (e) GDPNP-II complex. (f–h) Close-up views of EF-G domain IV interactions with the 30S subunit head in the (f) EF-G-post complex; (g) GDPNP-I and Fus complex; and (h) GDPNP-II complex. 16S rRNA, cyan; 30S proteins, blue; 23S rRNA, grey; 50S proteins, magenta; mRNA, green; P/P tRNA, red; E/E tRNA, yellow; pe∗/E tRNA, red; EF-G, orange. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 311

The Molecular Mechanics of the Ribosome 311

To determine the structures of rotated translocation intermediates, we trapped the ribosome either with fusidic acid (which allows hydrolysis of GTP but prevents release of EF-G) or the non-hydrolyzable GTP analogue GDPNP. Three diﬀerent

trapped complexes were crystallized (one with fusidic acid and two with GDPNP) ˚ ˚ and their structures solved at resolutions between 3.5 Aand4.1A (Fig. 2; [6]). These structures provide numerous insights into the mechanism of translation, including conformational changes that take place in the ribosome, tRNA and EF-G, as well as details of the many changes in intermolecular contacts that occur during translocation. The fusidic acid complex and one of the GDPNP complexes (GDPNP-I) shows 30S head rotation of 15◦, while the GDPNP-II complex shows a

distribution is allowed. distribution ◦ head rotation value of 18 , similar to that observed in the Spahn cryo-EM recon-

struction [5]. The tRNA is trapped in intermediate states between the P/E and E/E states (the pe∗/E state), in which the anticodon stem-loop (ASL) of the tRNA 9180 is bound to the P site of the 30S head, but also to a position near the E site of the only. No further No only. 30S body. (We term these states “chimeric hybrid states” following the terminology of Greek mythology in which the beast called chimera has the head of a lion and the body of a goat.) Most intriguingly, we observe that two universally conserved bases of 16S rRNA intercalate between bases of the mRNA, potentially acting as “pawls”

vayinstitutes.be/ of a translocational ratchet. The conserved and functionally important “switch loop PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- I” in the GTPase active site of EF-G is found to be ordered in the GDPNP structures for the ﬁrst time. Comparison with structure of the fusidic acid-stabilized complex suggests that fusidic acid acts by mimicking the structure of the ordered switch http://www.sol loop I. host on on host A working model for the mechanism of translocation Our ﬁndings, together with previous results from many other laboratories, suggest a model for translocation involving a sequence of large-scale structural changes in the ribosome. (i) Following peptide bond formation, the tRNAs move into the A/P http://www.worldscientific.com/worldscibooks/10.1142/ and P/E hybrid states. This movement is accompanied by forward (counterclockwise) intersubunit rotation, and is promoted by, but not necessarily dependent on, binding of EF-G·GTP. (ii) Forward (counterclockwise) rotation of the head of the 30S subunit advances the mRNA and the ASLs of the tRNAs and their associated mRNA codons relative to the body of the 30S subunit, into the ap/P and pe/E chimeric hybrid states. This step is dependent on binding EF-G, but does not ap-

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World pear to require GTP hydrolysis. (iii) Reverse (clockwise) rotation of the 30S head while disengaged from the ASLs completes movement of the tRNAs into their P/P NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY and E/E states. This may be accompanied by, or followed by, reverse (clockwise) rotation of the 30S subunit body, returning the ribosome to its classical, non-rotated

conformation. Hydrolysis of GTP is required for release of EF-G, but also appears to be necessary for some other step(s) in the translocation process [9]. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 312

312 J. Zhou et al.

Outlook for the future of research on ribosome translocation In spite of decades of research, often using ingenious and powerful approaches, our understanding of the mechanism of ribosomal translocation remains incomplete.

Among the many remaining mysteries are: (i) Is the mRNA moved actively or passively during translocation? (ii) How are mRNA and tRNA movement coupled, so as to prevent frame-shifting? (iii) What is the role of EF-G in translocation? (iv) What does GTP hydrolysis accomplish? (v) How can EF-G bind when the A and P sites are occupied by tRNA? (vi) What is “the ratchet”? (vii) What is the pawl of the ratchet? (viii) How are the movements of the 30S body and head, and other dynamic features, coordinated? distribution is allowed. distribution

Acknowledgments 9180 We thank the U.S. National Institutes of Health, the Agouron Foundation and the only. No further No only. California Institute for Quantitative Biosciences (QB3) for support of this work.

References

vayinstitutes.be/ 1. J. M. Ogle, D. E. Brodersen, W. M. Clemons Jr., M. J. Tarry, A. P. Carter PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

et al., Science 292, 897 (2001). - 2.P.Nissen,J.Hansen,N.Ban,P.B.Moore,T.A.Steitz,Science 289, 920 (2000).

http://www.sol 3. D. Moazed, H. F. Noller, Nature 342, 142 (1989). 4. J. Frank, R. K. Agrawal, Nature 406, 318 (2000).

host on on host 5.A.H.Ratje,J.Loerke,A.Mikolajka,M.Br¨unner, P. W. Hildebrand, et al., Nature 468, 713 (2010). 6. J. Zhu, A. Korostelev, D. A. Costantino, J. P. Donohue, H. F. Noller et al., Proc. Natl. Acad. Sci. USA 108, 1839 (2010).

http://www.worldscientific.com/worldscibooks/10.1142/ 7. Z. Guo, H. F. Noller, Proc. Natl. Acad. Sci. USA 109, 20391 (2012). 8. D. N. Ermolenko, H. F. Noller, Nat. Struct. Mol. Biol. 18, 457 (2011). 9. M. V. Rodnina, A. Savelsbergh, V. I. Katunin, W. Wintermeyer, Nature 385, 37 (1997). 10. Y. G. Gao, M. Selmer, C. M. Dunham, A. Weixlbaumer, A. C. Kelley et al., Science 326, 694 (2009). ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 313

313

EXPLORING THE DYNAMICS OF SUPRAMOLECULAR MACHINES WITH CRYO-ELECTRON MICROSCOPY

JOACHIM FRANK Howard Hughes Medical Institute, Department of Biochemistry and Molecular Biophysics 650 W. 168 th Street, New York, NY 10032, USA distribution is allowed. distribution

Present state of research 9180 Supramolecular machines perform their work in the cell by going through many dif- only. No further No only.

ferent states, distinguished by diﬀerent conformations and free-energy levels. Ide- ally, in order to ﬁnd out how these machines work, we would create a suitable in vitro environment containing all components including energy supply that allows the machine to function. We would then aim to take a “movie,” capturing their

vayinstitutes.be/ structure at highest resolution in a continuous fashion. Keeping within that ﬁlm PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

analogy, we might consider taking a large number of “snapshots” in equal small - time intervals, each short enough, as in the macroscopic world, to eliminate jarring transitions. However, we would ﬁnd out that this project has ﬂaws both on the

http://www.sol conceptual and the practical level. Conceptually, it is incorrect to equate a molecular machine’s progress to the workings of a macroscopic machine in motion since

host on on host the states are not ordered in sequence of time but are visited in a stochastic manner, with occasional irreversible events such as NTP hydrolysis as the only mark of progress. In practical terms, there are in fact two problems, one aﬀecting the way data for any given state can be captured, the other aﬀecting the ability to obtain

http://www.worldscientific.com/worldscibooks/10.1142/ coverage of states in a continuum. First of all, the visualization of a structure requires some form of radiation which imparts energy on the molecule and changes it in the very process. Minimization of these damaging interactions calls for a radiation dose so low that averaging over a suﬃciently large population is required for visualization. This requirement, in turn, translates into a complicated way each snapshot must be taken: the structural information in every state has to be gathered from many diﬀerent copies of

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World the molecule. While in forming such an average, crystallographic approaches — X-ray and electron crystallography — are able to take advantage of the regular NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY arrangement of molecules in a crystal, single-particle electron microscopy must ﬁrst determine the precise orientation of each molecule from its projection image [1].

The second hurdle interfering with the idea of making a movie is that only a limited number of states are suﬃciently populated to allow a three-dimensional structure to be determined. Thus the “movie frames” in between these states remain empty, undetermined. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 314

314 J. Frank

The diﬀerence between crystallographic and single-molecule approaches has opposite consequences for resolution and functional relevance: the high order achievable in a crystal makes it possible to obtain very high resolution, but the confor-

mational state that the molecule is trapped in may not be relevant to its function.

When, on the other hand, the structure is obtained from multiple images of free- standing “single” molecules as they are engaged in their work, functional relevance is guaranteed (e.g., [2]), yet until recently atomic resolution has not been achieved, except for viruses and other molecules with high symmetry [3, 4]. However, the introduction, in the past year, of direct-detection cameras [5, 6], some of which have single-electron counting and super-resolution capabilities, has distribution is allowed. distribution radically changed this situation. Recent accomplishments obtained with the help

of such cameras portray a field in rapid transition, with the ultimate claim to occupy a position in structural biology matching the one held until now by X- 9180 ray crystallography. Three-dimensional density maps (reflecting the reconstructed only. No further No only. Coulomb potential in a 3D array) obtained in the 3 A˚ range for particles with high symmetry [7, 8] and now even entirely asymmetric assemblies [9] enable de novo chain tracing and the construction of accurate atomic models without the aid of fitting existing structures. This of course pertains not only to the protein parts of

vayinstitutes.be/ a molecular machine, but to nucleic acid components, as well. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Another aspect to be emphasized in single-particle electron microscopy is the ability to obtain an entire inventory of co-existing states of a macromolecule from a single sample. Recent development of powerful software using maximum-likelihood http://www.sol methods has made it possible to extract multiple structures, one for each state, from a heterogeneous mixture [10]. This capability means that to some extent, with help host on on host from other techniques such as single-molecule FRET, inferences can be drawn on the dynamics of the supramolecular machine (“Story in a Sample” [11]). So far I have spoken of a reductionist, in vitro approach by electron microscopy toward study of supramolecular machines, whose aim is a portrayal of their structure

http://www.worldscientific.com/worldscibooks/10.1142/ and dynamics at atomic resolution. Another approach, electron tomography,isthe attempt to visualize the machine within the context of the cell. The progress with this approach in recent years, with automated tilt data collection having become routine, has been quite impressive [12]. In rare cases the interesting part of a cell is thin enough to be penetrated by the electron beam. For thicker samples, high-pressure freezing must be used followed by sectioning with diamond knife or Focused Ion Beam (FIB) milling. While FIB milling of a frozen sample requires

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World very specialized equipment, it is becoming the preferred method of sectioning of

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY biological material as it is virtually free of artifacts associated with cutting [13, 14].

My lab’s recent research contributions

My lab studies the mechanism of protein biosynthesis in both eubacteria and eukaryotes, using single-particle cryo-electron microscopy of ribosomes that are in various states of translation. These states are characterized by diﬀerent conformations of October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 315

Exploring the Dynamics of Supramolecular Machines with Cryo-Electron Microscopy 315

the ribosome itself and diﬀerent binding conﬁgurations of mRNA, tRNA, as well as a variety of translation factors (in the case of bacteria, EF-G, EF-Tu, etc.). From density maps reconstructed, such as in ref. [2], atomic models are built by dock-

ing and ﬂexible ﬁtting of structures in the Protein Data Bank. Of special interest

to us has been translocation, or the mechanism by which the ribosome transports mRNA and tRNAs bound to it to the next codon in each cycle of the polypeptide elongation. We early recognized that this movement is facilitated by a ratchet- like reorganization of the ribosome [15]. More recently we discovered [16] that the ribosome at the point after peptide bond formation exists in multiple conformational states characterized by different intersubunit rotations and tRNA binding distribution is allowed. distribution configurations. Thanks to the above-mentioned development of novel classification

software these states could be extracted and independently reconstructed [17]. Most recent contributions by my group to the ﬁeld of biosynthesis include the 5-A˚ 9180 resolution reconstruction of the ribosome from Trypanosoma brucei,aeukaryotic only. No further No only. parasite causing Sleeping Sickness [18], and the determination of the structure of the mammalian translation pre-initiation complex [19]. At the date of preparation of this manuscript, the T. brucei ribosome reconstruction represented one of the highest-resolution structures of an asymmetric molecule, though still employing the

vayinstitutes.be/ old technology of recording on ﬁlm (Fig. 1). Compared to the ribosomes of other PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- eukaryotes, the T. brucei ribosome possesses unusual features, such as its large RNA expansion segments which may be docking platforms for T. brucei-speciﬁc protein factors, possibly associated with the need of the parasite to rapidly adapt to hosts http://www.sol with widely diﬀerent body temperatures. host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Fig. 1. Cryo-EM density map of the ribosome from Trypanosoma brucei at 5 A˚ resolution. (Left) The ribosome (grey) viewed from the solvent side of the small subunit; (right) viewed from the solvent side of the large subunit. Ribosomal RNA expansion segments are rendered in diﬀerent colors. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 316

316 J. Frank

The visualization of the mammalian 43S pre-initiation complex presents an example for a highly heterogeneous specimen that could nevertheless be characterized by a series of reconstructions which depict diﬀerent combinations of factors attached

to the small ribosomal subunit. Of these, only one representing 4.4% of the whole

dataset contained the initiator tRNA and most factors that are important in setting up the 43S initiation complex and the staging for the scanning of the mRNA for the start codon (Fig. 2). distribution is allowed. distribution

9180 only. No further No only.

˚ vayinstitutes.be/ Fig. 2. 43S translation pre-initiation complex at 11.6 A resolution, reconstructed from a subpopulation of 29,000 cryo-EM particle images. The components of this supramolecular complex are: PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 40S ribosomal subunit (yellow), eIF3 (red), DHX29 (green), initiator tRNA (yellow-orange), and eIF2 (orange, attached to initiator tRNA). http://www.sol Outlook to future developments of research

host on on host As noted in the introduction, the new generation of detectors employed in the transmission electron microscope is currently revolutionizing cryo-electron microscopy as a tool in structural research, in particular in the study of supramolecular complexes. Due the gain in contrast and resolution achievable, it will soon be possible

http://www.worldscientific.com/worldscibooks/10.1142/ to reconstruct supramolecular machines in their entirety — proteins along with the nucleic acid components — in multiple, functionally relevant states. The novel technology has a bearing on both diﬀerent experimental approaches outlined in the beginning: single-particle reconstruction and electron tomography of cell sections. Both techniques will be transformed because of the gains in resolution. Determi- nation of atomic structures by single-particle reconstruction will become routine for “well-behaved” supramolecular complexes, i.e., those that occur only in a few

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World well-populated states. High-throughput methods will be available similar to those now found in X-ray crystallography. At the same time, electron tomography will NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY reach the range of resolutions that allows the signatures of conformational states of a molecule to be recognized within the context of a cell, enabling the tracking

of information relevant for the description of its functional state in situ.Inthis way, it will be possible to directly link localized processes in the cell to the dynamical behavior of atomic structures determined by a combination of single-particle cryo-EM, X-ray crystallography, and single-molecule FRET. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 317

Exploring the Dynamics of Supramolecular Machines with Cryo-Electron Microscopy 317

Acknowledgments Funding has been provided by Howard Hughes Medical Institute and grants NIH R01 GM29169 and GM55440. I thank Yaser Hashem for the preparation of the

ﬁgures and for a critical reading of the manuscript.

References 1. J. Frank, Three-dimensional Electron Microscopy of Macromolecular Assem- blies (Oxford University Press, 2006). 2. E. Villa, J. Sengupta, L. G. Trabuco, J. LeBarron, W. T. Baxter et al., Proc. distribution is allowed. distribution Natl. Acad. Sci. USA 106, 1063 (2009).

3.X.Zhang,L.Jin,Q.Fang,W.H.Hui,Z.H.Zhou,Cell 141, 472 (2010).

9180 4. N. Grigorieﬀ, S. Harrison, Curr. Opin. Struct. Biol. 21, 265 (2011). 5. A. R. Faruqi, J. Phys. Condens. Matter 21, 314004 (2009). only. No further No only. 6.X.Li,S.Zheng,C.R.Booth,M.B.Braunfeld,S.Gubbenset al., Nat. Meth. 10, 584 (2013). 7. B. E. Bammes, R. H. Rochat, J. Jakana, D.-H. Chen, W. Chiu, J. Struct. Biol. 177, 589 (2012).

vayinstitutes.be/ 8. D. Veesler, T.-S. Ng, A. K. Sendamarai, B. J. Eilers, C. M. Lawrence et al., Proc. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Natl. Acad. Sci. USA 110, 5504 (2013). 9. X.-C. Bai, I. S. Fernandez, G. McMullan, S. H. W. Scheres, eLife 2:e00461 (2013). http://www.sol 10. S. H. Scheres, J. Mol. Biol. 415, 406 (2012). 11. J. Frank, Biopolymers 99, 832 (2013). host on on host 12. V. Luˇci´e, A. Rigort, W. Baumeister, J. Cell Biol. 202, 407 (2013). 13. M. Marko, C. Hsieh, R. Schalek, J. Frank, C. Mannella, Nat. Meth. 4, 215 (2007). 14. K. Wang, K. Strunk, G. Zhao, J. L. Gray, P. Zhang, J. Struct. Biol. 180, 318

http://www.worldscientific.com/worldscibooks/10.1142/ (2012). 15. J. Frank, R. K. Agrawal, Nature 406, 318 (2000). 16. X. Agirrezabala, J. Lei, J. L. Brunelle, R. F. Ortiz-Meoz, R. R. Green et al., Mol. Cell 32, 190 (2008). 17. X. Agirrezabala, H. Liao, E. Schreiner, J. Fu, R. F. Ortiz-Meoz et al., Proc. Natl. Acad. Sci. USA 109, 6094 (2012). 18. Y. Hashem, A. des Georges, J. Fu, S. N. Buss, F. Jossinet et al., Nature 494, ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 385 (2013).

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 19. Y. Hashem, A. des Georges, V. Dhote, R. Langlois, H. L. Liao et al., Cell 153, 1108 (2013).

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 318

318

CRYSTALLOGRAPHIC STUDIES OF EUKARYOTIC RIBOSOMES AND FUNCTIONAL INSIGHTS

NENAD BAN Institute of Molecular Biology and Biophysics, ETH Z¨urich (Swiss Federal Institute of Technology), Z¨urich, 8093, Switzerland distribution is allowed. distribution

State of research (proteins in supremolecular machines) 9180 Recent developments in the technology and methods for crystallographic and elec- only. No further No only. tron microscopic data collection and structure determination have made it possible to approach particularly complex problems in structural biology [1, 2]. Some of these improvements include; i) the development of methods to rapidly freeze samples in the native hydrated state for crystallographic and electron microscopic experiments,

vayinstitutes.be/ ii) the availability of high-intensity synchrotrons and beamlines with a tunable wave- PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- length for crystallographic experiments and the availability of a new generation of very stable electron microscopes with possibilities for automatic data acquisition, iii) the use of extremely sensitive direct detection devices, iv) improved software for http://www.sol data integration and phasing of the measured amplitudes in crystallography and for particle alignment and 3D reconstruction in cryo-electron microscopy combined host on on host with an increase in computer speed [3]. These developments are now propelling the studies of large macromolecular assemblies to a new level and several new structures have emerged in the past few years that reveal interesting architectural features of protein components in large cellular complexes and their functional roles, for some

http://www.worldscientific.com/worldscibooks/10.1142/ examples see: [1, 4–7]. The section below outlines our contributions to the ﬁeld using as an example our structural studies of eukaryotic ribosomes, multi-megadalton ribonucleoprotein complexes responsible for protein synthesis.

Recent research contributions (proteins in supramolecular machines)

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World The ribosome is a large cellular assembly that plays a central role in the process of protein synthesis in all organisms [8]. Although basic aspects of protein synthesis are NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY preserved in all kingdoms of life, eukaryotic ribosomes are much more complex than their bacterial counterparts, require a large number of assembly and maturation

factors during their biogenesis, use numerous initiation factors, and are subjected to extensive regulation [9, 10]. Recently we obtained detailed structural information on eukaryotic ribosomes by determining the ﬁrst complete structures of both eukaryotic ribosomal subunits October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 319

Crystallographic Studies of Eukaryotic Ribosomes and Functional Insights 319

each in complex with an initiation factor [9, 11]. These results are of fundamental importance for understanding protein synthesis in eukaryotes and are the prerequisite for understanding the structural basis of the regulation of protein synthesis in

normal cells and how it is perturbed in various diseases.

Crystal structure of the eukaryotic 40S ribosomal subunit in complex with initiation factors 1 and 1A The eukaryotic 40S ribosomal subunit plays a central role in the initiation of protein synthesis during which it interacts with more than a dozen eukaryotic translation distribution is allowed. distribution

initiation factors (three in bacteria) [12]. These functional diﬀerences are reﬂected

in the size and complexity of the eukaryotic 40S subunit; it is 500 kDa larger than

9180 the 30S subunit with 18 eukaryotic proteins that have no homologues in bacteria and it contains a considerably longer rRNA. Furthermore, due to its role in transla- only. No further No only. tion initiation, which diﬀers substantially between prokaryotes and eukaryotes, the eukaryotic 40S subunit is the target of regulation in a number of cellular processes including development, diﬀerentiation, stress response and neuronal function. vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

Fig. 1. Two views of the small 40S subunit of the eukaryotic ribosome. Proteins in this ribonucleoprotein complex the colored based on conservation in bacteria, archaea and eukaryotes. Structures of individual proteins forming the complex are shown as a collage on the right. Their evolutionary origin is indicated by the coloring scheme. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 320

320 N. Ban

To gain insight into the functioning of the eukaryotic ribosome and the process of translation initiation in eukaryotes, we determined the crystal structure of the eukaryotic 40S ribosomal subunit in complex with eukaryotic initiation factors 1

and 1A [5, 13]. The structure deﬁnes the locations and the folds of all 33 eukaryotic

ribosomal proteins (Fig. 1) and reveals the fold of the 18S rRNA. Furthermore, it provides information on proteins responsible for ribosomal functions in eukaryotic- speciﬁc signaling and regulation, and it permits mapping of mutations that lead to genetic diseases. Finally, the interactions with the two initiation factors eIF1 and eIF1A provide a basis for understanding their roles during eukaryotic translation initiation and the role of mutations that aﬀect the start codon recognition [14]. distribution is allowed. distribution

Crystal structure of the eukaryotic 60S ribosomal subunit in

9180 complex with initiation factor 6 The eukaryotic 60S ribosomal subunit catalyzes peptide bond formation and pro- only. No further No only.

vides a platform for binding of various factors involved in nascent polypeptide processing, folding, targeting to the membranes and membrane insertion. The 60S subunit is considerably larger and more complex than its bacterial counterpart [15], with many additional eukaryotic-speciﬁc functions connected to the regulation of vayinstitutes.be/ protein synthesis or ribosome assembly and maturation. Furthermore, the large PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- ribosomal subunit is a major target for various antibiotics, and therefore revealing http://www.sol host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Fig. 2. Two views of the 60S subunit of the eukaryotic ribosome. Proteins in this ribonucleoprotein complex are colored based on conservation in bacteria, archaea and eukaryotes. Structures of individual proteins forming the complex are shown as a collage on the right. Their evolutionary origin is indicated by the coloring scheme. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 321

Crystallographic Studies of Eukaryotic Ribosomes and Functional Insights 321

the differences between the eukaryotic 60S subunit and the bacterial 50S subunit is important for our understanding of antibiotic specificity of compounds that target the bacterial ribosome and those that specifically inhibit the eukaryotic ribosome.

Recently we described a complete atomic structure of the 60S ribosomal subunit

in complex with the eukaryotic initiation factor 6 [6]. The structure of the eukaryotic large ribosomal subunit reveals a network of interactions between eukaryotic- specific ribosomal proteins and ribosomal RNA expansion segments, demonstrating that novel eukaryotic-specific architectural solutions provide long-distance tertiary stabilization of this ribonucleoprotein complex (Fig. 2). It uncovers the roles of eukaryotic ribosomal protein elements in the stabilization of the active site and the distribution is allowed. distribution extent of eukaryotic-specific differences in other functional regions of the subunit.

The structure also offers insights into the regulatory mechanisms of the eukaryotic ribosome, the maturation of the large ribosomal subunit, the principles of 9180 antibiotic specificity, and the structural basis of extra-ribosomal roles of various only. No further No only. eukaryotic-specific ribosomal proteins. Finally, it elucidates the molecular basis of the interaction with eukaryotic initiation factor 6, involved in both initiation of protein synthesis and ribosome maturation; The release of eIF6 from the ribosome has been connected with the Shwachman-Bodian-Diamond syndrome, an inherited

vayinstitutes.be/ disorder with a wide range of abnormalities and symptoms. PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

Outlook and future developments of research (proteins in

http://www.sol supramolecular machines) Based on the established methodology and obtained structural information we are

host on on host now pursuing studies of various functional complexes of eukaryotic, archaeal and bacterial ribosomes to better understand initiation of translation in eukaryotes and ribosomal evolution and assembly in general. Future developments in this ﬁeld are likely to stem from the above described developments in the ﬁeld of cryo-electron

http://www.worldscientific.com/worldscibooks/10.1142/ microscopy combined with single particle reconstruction techniques. This is emerging as one of the most powerful methods in structural biology with huge potential for future development in the following years. Over the next few years atomic resolution reconstructions using cryo-electron microscopy will be in reach for a number of interesting biological systems.

Acknowledgments ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World The author thanks F. Voigts-Hoffmann for generating the figures, D. Boehringer, S. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Klinge, M. Leibundgut, F. Voigts-Hoffmann and J. Rabl for discussions. Research described here was supported by the Swiss National Science Foundation (SNSF),

the National Center of Excellence in Research (NCCR) Structural Biology program of the SNSF, and European Research Council grant 250071 under the European Community’s Seventh Framework Programme (N.B.), and the Max Roessler prize to N.B. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 322

322 N. Ban

References 1. X. C. Bai, I. S. Fernandez, G. McMullan, S. H. Scheres, eLife 2, e00461 (2013). 2. M. Mueller, S. Jenni, N. Ban, Curr. Opin. Struct. Biol. 17, 579 (2007).

3. J. P. Abrahams, N. Ban, Meth. Enzymol. 374, 163 (2003). 4. A. Ben-Shem, N. Garreau de Loubresse, S. Melnikov, L. Jenner, G. Yusupova et al., Science 334, 1524 (2011). 5. J. Rabl, M. Leibundgut, S. F. Ataide, A. Haag, N. Ban, Science 331, 730 (2011). 6. S. Klinge, F. Voigts-Hoﬀmann, M. Leibundgut, S. Arpagaus, N. Ban, Science 334, 941 (2011). distribution is allowed. distribution 7. D. L. Makino, M. Baumgartner, E. Conti, Nature 495, 70 (2013). 8. T. M. Schmeing, V. Ramakrishnan, Nature 461, 1234 (2009).

9180 9. S. Klinge, F. Voigts-Hoﬀmann, M. Leibundgut, N. Ban, Trends Biochem. Sci. 37, 189 (2012). only. No further No only. 10. S. Melnikov, A. Ben-Shem, N. Garreau de Loubresse, L. Jenner, G. Yusupova et al., Nat. Struct. Mol. Biol. 19, 560 (2012). 11. F. Voigts-Hoﬀmann, S. Klinge, N. Ban, Curr. Opin. Struct. Biol. 22, 768 (2012). vayinstitutes.be/ 12.J.R.Lorsch,T.E.Dever,J. Biol. Chem. 285, 21203 (2010). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- 13. M. Weisser, F. Voigts-Hoﬀmann, J. Rabl, M. Leibundgut, N. Ban, Nat. Struct. Mol. Biol. 20, 1015 (2013). 14. C. E. Aitken, J. R. Lorsch, Nat. Struct. Mol. Biol. 19, 568 (2012). http://www.sol 15. N. Ban, P. Nissen, J. Hansen, P. B. Moore, T. A. Steitz, Science 289, 905 (2000). host on on host http://www.worldscientific.com/worldscibooks/10.1142/ ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 323

323

SESSION 6: PROTEINS IN SUPRAMOLECULAR MACHINES

CHAIR: RUDOLF GLOCKSHUBER AUDITORS: J.-F. COLLET1,H.REMAUT2 (1) Welbio, Avenue Hippocrate 75, 1200 Brussels, Belgium and de Duve Institute, Universit´e catholique de Louvain, Avenue Hippocrate 75, 1200 Brussels, Belgium (2) Structural and Molecular Microbiology, VIB Department of Structural Biology, VIB, distribution is allowed. distribution and Structural Biology Brussels, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium

9180 Discussion among panel members only. No further No only.

Rudi Glockshuber: I have a question to Craig Townsend. When you do this polyketide synthesis in vitro, for example with puriﬁed polyproteins and all the substrates, what are the best yields, which have so far been reported? How many turnovers did you get and how pure is the end-product that you get? vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Craig Townsend: The end-product is surprisingly pure. The interesting thing is that one can combine these proteins and they will actually still work together despite

http://www.sol having been taken apart. They even faithfully do the chemistry that you expect them to do. I think that one of the disadvantages of taking apart, deconstructing

host on on host them, is that the efficiency of the process is worse probably than when all proteins are together. So this is an important fundamental question and we have thought about it a lot. The amounts we are making are small, but I think that the comparison that will be useful to make is to have a fully intact protein and measure an overall flux without and with the various components. We have done a combination http://www.worldscientific.com/worldscibooks/10.1142/ of two parts of three domains apiece, we have done three plus two plus one and three plus one plus one plus one. We can see just qualitatively by HPLC that the efficiency drops down. I think it can be reasonably efficient.

Rudi Glockshuber: And what is the maximum number of turnovers which has been observed?

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Craig Townsend: I don’t know the answer yet but I think we can run these reactions

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY over periods of hours and still see production. I think the proteins are fairly robust if you cut them correctly, if you disassemble them correctly. The individual pieces are fairly stable and they will associate suﬃciently well to be pretty good little machines.

Nenad Ban: I have a question for Ian Wilson. I noticed that with the HIV trimer structural interpretation, that you have been able to obtain at about 5 A,˚ you had, October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 324

324 Proteins in Supramolecular Machines

sometimes, atomic models underlying this structure. The question is: it is in a way very tricky to make interpretations at 5 A,˚ so I presume that in some cases you had previously solved domains. But do you think that, in some cases, you can actually

trace the fold? We have done a lot of interpretations at this resolution range and

it seems extremely diﬃcult, let say, to either use molecular replacement procedures or perform very accurate fold traces at that resolution.

Ian Wilson: We have a lot of models for about 50% of this structure. And we also have the model of the antibody. So the antibody can be placed really pretty accurately. We have a large part of the top domain, which is gp120. So it is missing distribution is allowed. distribution

some of the hypervariable loops. However, there has been multiple structures solved

of that, and so they can be really ﬁtted quite well in the structure solved. For the molecular replacement, we have about 50% of the mass that we were using to get the 9180 solution. We haven’t a lot of the pieces that we did not have for helices, which we ˚ only. No further No only. can see quite well below 5 A. And so we have been fairly conservative about building

in only the helical regions. We have density, which can account for all of the rest of it but we are cautious about the interpretation of that at the moment. We have both the cryo-EM structure and the X-ray structure and we do not have to deal with molecular replacement for the cryo-EM structure. These were independently vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS solved and, basically, all of the main features, all of the main new features agreed,

- as well as the ﬁtting of the previously known structures.

http://www.sol Joachim Frank: This is a follow-up question (to Ian Wilson). You have this 5.7 A structure. How does it compare to the controversial structure that was published

host on on host by Mao et al. in PNAS who claimed a very similar resolution?

Ian Wilson: There is almost no correlation at all of the new regions. Similar to what you described, by cryo-EM, we have multiple states in the reconstruction, in the sample. Andrew Ward’s group used an antibody for which there were some http://www.worldscientific.com/worldscibooks/10.1142/ unliganded complexes, one-liganded, two-liganded and three-liganded. They have done a reconstruction of each of these. We have looked at the limit at which we can actually do a reconstruction for the unliganded trimer and I think the best you can do without adding features such antibodies is around 12 A.˚ That would be very discrepant with the reported 6 A˚ structure. We have chosen not to do any comparison with that structure because I think it is highly controversial, but we see no features in common with our structure at all. And it does not look like a type I ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World membrane protein. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Rudi Glockshuber: I have also a question to Ian Wilson. How heterogeneous is your HIV protein with respect to the glycans?

Ian Wilson: So for the crystallography, (note that there are 81 glycans on the structure), we basically expressed them in cells that produce high mannose. We then bind the antibody. The antibody binds with high aﬃnity. And so we could October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 325

Session 6: Discussions 325

actually take PNGase F and actually clean oﬀ the rest of it down to single sugars. So that really enabled us to actually get a very nice homogenous preparation that could crystallize and get us to that resolution. For the EM, it is actually much

more interesting because we can actually express the glycans in 293T cells. And so

we actually have much more native-like glycans and those glycans can actually be observed in the EM density. Clearly there will be some heterogeneity there but we can already see at least the core of the glycans, which is the same for almost all of the glycans, appearing in the EM density. So this is really very exciting because you are not able to really deﬁne the epitopes unless you actually have the problem of glycosylation on there because many of these antibodies for example that we are distribution is allowed. distribution actually looking at require binding to both high mannose and to complex glycans.

So when one looks at the epitopes as deﬁned on the trimer structure as distinct from engineered smaller constructs, one actually ﬁnds that the epitopes are much 9180 larger, includes more glycans, and includes more surface, including the protomers, only. No further No only. adjacent protomers, that you don’t see when you are looking at monomer units.

Marina Rodnina: I have a question to Rudi Glockshuber. You have this very high aﬃnity, incredibly high, right? Can you understand that in molecular terms, so trace that high aﬃnity back to hydrogen bonds, to salt-bridges and so on? vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Rudi Glockshuber: The very slow oﬀ-rate is most likely caused by simultaneous formation of twenty inter-molecular β-sheet hydrogen bonds, because the bound

http://www.sol peptide is inserted between the neighboring strands of the protein and it is a con- tiguous hydrogen bond network of twenty simultaneous hydrogen bonds which you

host on on host would have to break to release the peptide. The complex really behaves like a single polypeptide chain. The peptide only falls oﬀ when the protein unfolds simultaneously. In addition, every second residue points towards the inside, because it is a β-strand, and there is also very good surface complementarity.

http://www.worldscientific.com/worldscibooks/10.1142/ Marina Rodnina: Isn’t it a problem for drug design because then you have to ﬁnd drugs which are competing with this extremely tight binding?

Rudi Glockshuber: I personally believe that pilus assembly inhibitors are not a good idea. The main reason is not so much the stability but rather that an assembly inhibitor does not destroy the pili which are already present when an infection has started. I think that a much more promising strategy is towards adhesion inhibitors

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World which compete, like mannose analogs for example, with receptors for binding of the

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY pilus, in combination with antibiotics therapy.

Nenad Ban: Another question for Rudi Glockshuber. Now that you are saying this,

it reminds me of the fact that, maybe there is nothing to it, but people sometimes say you should drink cranberry juice when you have an urinary tract infection. Could it be that in cranberry juice there is some compounds that actually might be mannose analogs or something like this? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 326

326 Proteins in Supramolecular Machines

Rudi Glockshuber: Yes, this is definitively true. I think some of the compounds have even been identified. There is another quite interesting aspect of the pili because they increase apparent affinity during sheer forces. For example, under

ﬂow conditions during excretion of the urine, they increase their apparent aﬃnity.

The reason is because there is most likely a domain separation between the pilin domain and the lectin domain under sheer forces. And this induces a conformational change in the lectin domain of the adhesin, which has a higher affinity and about five orders of magnitude lower off rate.

Rudi Glockshuber: If there are no further questions to the ﬁrst three talks, I would distribution is allowed. distribution

like to suggest that we maybe start quickly with a couple of questions on the ribo-

some talks and then make a break and continue afterwards.

9180 Marina Rodnina: I have a question to Joachim Frank. Which is about the time- resolved cryo-EM, so where are we there? only. No further No only.

Joachim Frank: I didn’t have an opportunity to talk about the experimental design. But there is a technology that employs a microchip with two inlets in a reaction chamber and then the compound it forms is spread directly on the grid, and the

vayinstitutes.be/ grid is plunged down into the cryogen. So one is able to get a time resolution down PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- to ten milliseconds with this device. We tested this device, which was manufactured at RPI, using ribosome subunit association as a sort of test case. We were able to identify through diﬀerent time points the formation of bridges, so we have http://www.sol intermediate states in the formation of bridges. So the entire thing really works and calls for applications that target kinetics of, for instance, the decoding process and host on on host so on.

Rudi Glockshuber: I also have a question to Joachim Frank. How dangerous, how big is the likelihood that you identiﬁed too many conformations? I mean some

http://www.worldscientific.com/worldscibooks/10.1142/ conformations that are maybe not physiologically relevant.

Joachim Frank: Yes, very dangerous. I think one really has to analyze the crop that comes from the classification very, very carefully. What we do, at the resolution that we achieved, we can really identify individual factors by fitting X-ray structures convincingly. I think there is a certain resolution threshold that has to be reached before one can confidently do this and below that threshold (i.e., when the resolution

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World is not good enough) one simply cannot.

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Nenad Ban: Harry, I have a question for you (Harry Noller). I think it is very beautiful the way you described it and the results that you’ve achieved of under-

standing the rotation of the head of the small subunit with respect to repositioning of the messenger RNA and the tRNAs and I remember this is diﬀerent from what was initially thought based on some early structures of the small subunit where, I think, it was frequently shown that the head has to open completely to let the October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 327

Session 6: Discussions 327

tRNA and the codon stem loop to move from P to E site and from A to P site. In your current understanding of how this repositioning occurs, you have stepwise events that include maintaining contacts with the head and then separately with

the body of the small subunit. Did you ever look at, let’s say, the extent of contacts,

or maybe, some energetic contributions, because obviously, this way you managed to break the contacts one at the time relative to the head or relative to the body. How would that agree with, let say, either the sequence of steps, or whether they are balanced, these aﬃnities?

Harry Noller: I think it is very early days still in this understanding of this process. distribution is allowed. distribution

For example we can see the rotation of the head and the tRNA following the rotation

of the head. But now we know the head has to return. So it must somehow release its contacts with the tRNA during the return. This would be some kind of ratchet 9180 mechanism where the pawl is released. I think we have no idea how this happens

only. No further No only. at present but it could involve something like what you ﬁrst mentioned, maybe an

orthogonal rotation of the head upwards, or maybe a deformation, for example, of helix 34 of 16S RNA that is sort of a scaﬀold for the things that contact the tRNA. So I think we are just at the beginning, we see the outlines ﬁnally of what the mechanism is for the two mains steps of translocation but the details that you are vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS eluding to, I think, are going to get to the essence of the mechanism, which again

- is going to be a RNA mechanism, no doubt.

http://www.sol Discussion among all attendees Rudi Glockshuber: I propose to continue our discussion on the speciﬁc issues on host on on host the ribosome talks and after that have a more general discussion on the main future challenges of structural biology of supra-molecular complexes.

Kurt W¨uthrich: I was very intrigued by the stops in the translation that you de-

http://www.worldscientific.com/worldscibooks/10.1142/ scribed. You alluded to the possibility that these are related to the folding of the nascent polypeptide chain. Do you have any evidence or even data to show for this?

Marina Rodnina: We don’t have our own evidence but there is evidence in the literature which suggest that if you mutate a rare codon of the mRNA to a more frequently used codon, there are cases that, thereby, you change the rate of translation and then you see apparently diﬀerent folding which is manifested through a

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World diﬀerent function of the protein. There is, in the meantime, I think, three or four reports in this direction. One of them was published, let’s say ﬁve years ago while NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY two or three are very recent. I have to say that so far we don’t have our own evidence and we are working on that. I also have to admit that these previous papers

have a number of issues. So it is not at all clear whether it is really the case or not. But there is an interesting indication, and I think this is an interesting way to go for. Also, this is deﬁnitively not true for all proteins, because from what we have seen for a number of proteins, for example, β-barrel proteins, which are very likely October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 328

328 Proteins in Supramolecular Machines

to fold post-translationally, they don’t have stops at all. But we’re just starting. So it is just the α-helical and the proteins of mixed α and β structures, which show these stops. Even we cannot exclude that this one particular topology which we

tend to study all the time, so this Rossmann-fold basically, makes all the stops. We

still have to see that.

Kurt W¨uthrich: Are the stops also related to proline or to multi-proline sequences, which deﬁnitively have a special role in protein folding?

Marina Rodnina: Not in these proteins. We have analyzed that very carefully; these proteins do not have double prolines. They have single prolines, which do not cause distribution is allowed. distribution

stopping. What we have seen is that there are some stops, which are deﬁnitively

due to rare codons and rare codon clusters. There are some stops, which are related

9180 to the Shine-Dalgarno-like sequences, very likely. But about 50% of stops which we have seen, we don’t know what is that. only. No further No only.

Nenad Ban: Because of these observations that Marina Rodnina mentioned, it was also suspected that, perhaps, this is partially the reason why occasionally eukaryotic proteins are diﬃcult to express in bacteria because if we optimize codon usage, then

vayinstitutes.be/ we get fast read-through through potential pause sequences which are important for PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- gradual domain-wise folding of large eukaryotic proteins. Now since it is possible to optimize codons for bacterial expression of eukaryotic proteins, it is also attempted to simulate a little bit the distribution of codon as it is observed in the eukaryotic http://www.sol system.

host on on host Jason Chin: Marina, I was curious about the EF-P experiment. There is an interesting thing going on there both with the modiﬁcation of EF-P and with what is happening mechanistically at the peptidyl-transferase center. Speciﬁcally, I was curious, if you like, on the substrate side, whether you have been able to dissect out

http://www.worldscientific.com/worldscibooks/10.1142/ when EF-P is necessary? If you do the experiment, can you do the experiment, for example, with N-methylated but non-cyclic amino acids to dissect out the eﬀect of N-methylation versus the geometry constraint? What do you think mechanistically is happening in the peptidyl-transferase center in turn on EF-P, do you actually think that there is facilitated RNA catalysis or whether the protein is participating directly in facilitating peptide-bond formation?

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Marina Rodnina: Good question and very little answers. First of all, it is diﬃcult to say how EF-P is contributing to catalysis because in the ﬁrst place we don’t NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY understand why proteins are so slow. Apparently, when two prolines are there, there is something in the structure of the catalytic center which is not positioned

properly. So why proline does not do that, we simply don’t know. Of course we also do not know what EF-P is doing. All these other experiments are all planned and should be done. The point is that, if you look at the exact contribution, of for example, modiﬁcation which is looking in the peptidyl-transferase center, this is October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 329

Session 6: Discussions 329

only a factor of four to catalysis. This might mean that there are two contributions. One of them is that EF-P is doing in fact something to the t-RNA, somehow pushing it into proper position and then there is additional catalytic eﬀect.

Sachdev Sidhu: What about relationship between pausing and secretion? Is there

any inﬂuence on what happens in the ER with glycosylation, etc.?

Marina Rodnina: This would be all in eukaryotic systems. We know that there is pausing. But this is usually not necessarily by the pausing which we are talking about during translation, because there is specialized pausing, for example translational arrest. I can imagine there are more tools there. I suppose this is type distribution is allowed. distribution

of pausing which is much slower than the pausing which I am talking about. One

should not forget that this pausing during translation introduces small pauses. They

9180 would not be sufficient to put it somewhere in a different compartment. only. No further No only. Sachdev Sidhu: When it is already going into the compartment is there influence on how it folds in the ER if something is stalled?

Marina Rodnina: This is a good question. We don’t know how this pausing is

vayinstitutes.be/ aﬀected by anything from the outside. Also not by all these proteins which are PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- interacting with the exit tunnel of the ribosome, and there are many of them (Nenad mentioned some of them). For example, we know that SRP, in bacteria, does not affect it. But in eukaryotes it does affect it, very strongly. Trigger factor does not http://www.sol seem to affect it, from what we know. But for the rest, we don’t know such as if targeting to the translocon affects the rate of translation, this is really not known. host on on host Chi-Huey Wong: It is exciting to see the combination of EM and X-ray. I have a question about glycosylation of membrane proteins. We know in eukaryotic system, it is co-translational. The question is how would the system do that? And how

http://www.worldscientific.com/worldscibooks/10.1142/ would the system ensure that after glycosylation and before folding, the sugar chain would be on the surface instead of the interior of the protein. Is there a correction system there to ensure this process?

Marina Rodnina: This is a very complex question, so I’ll try to give a very short answer to that. This is probably an interplay between many components in the system. The first step in this process will be targeting of the ribosome with the nascent peptide to the translocon. And then this peptide continues to be synthesized ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World while being in the translocon or starting to exit the translocon. These helices, which NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY are to be in the membrane at the end, are synthesized one by one. They have to find their proper topology. How they find their proper topology is not exactly clear.

There are models that this occurs spontaneously due to physical properties of the trans-membrane segments themselves and the membrane. But I don’t think we know more. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 330

330 Proteins in Supramolecular Machines

Ray Stevens: We are reaching the end of the meeting, so it is time for sort of crazy questions. Each of you showed these beautiful supra-molecular assemblies. The next level of complexity is the cell. When are we going to see sort of the structure

of the cell and what technology for prokaryotes and eukaryote? Any comments on

that?

Joachim Frank: Well there is, as you might know, technology available already with electron microscopy, so-called electron tomography. In this case, one actually actively tilts a cell in all possible directions and then forms a three-dimensional image. The application of this technology is severely restricted by the limited distribution is allowed. distribution

penetration range of the electrons, so one can only work with 0.2-0.3 microns. One

either has to use very thin protrusions of the cell or, if interested in the entire cell, one has to use some kind of slicing method. There is the “FIB” (Focused Ion Beam) 9180 milling technology which is now being increasingly applied in biology. It is possible

only. No further No only. to actually look at selected sections of a cell and reconstruct these in their entirety.

But of course the resolution is limited, and one does not get into the sub-nanometer range that we are now used to with the single-particle reconstruction methods.

Ray Stevens: Sort of a follow-up on that, it seems that one of the limiting factors vayinstitutes.be/ is going to be the hybrid approach. . . we are going to use multiple technologies PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- again, like each of you sort of demonstrated, but computational is probably going to be one of the biggest, I think, limitations to really advance this kind of dreams

http://www.sol or visions. Any comments on what computationally, from anybody else, what the breakthrough is going to be, to imagine this type of dream? host on on host Gebhard Schertler: Can I first say something to experimental? There is soft X-ray imaging which has made a lot of progress, particularly in Berlin and in the US. Recently I have seen a first tomogram where we can see actually tubulin subunits in a soft tomogram of a full prokaryotic cell. If you want to go to high resolution and http://www.worldscientific.com/worldscibooks/10.1142/ even thickest slices (here you can go to 10 microns of range in one tomogram), you have to use hard X-rays. You can then go to 50 microns most likely. That covers roughly a cell or a cell-cell interphase. In the very long run, this is not the very close future, the hard X-ray imaging will come along with a few techniques. I think it will fill in a range above the EM tomography. When we can look for example at membrane topology in very great details and then go into a certain section and do an EM tomogram. In this way, we kind of close a blind gap which is now very ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World difficult to access. It might also become faster than, for example, these methods at

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY the moment where we do block phase imaging, which are roughly doing the same, but very slowly.

Chris Walsh: Given that the questions changed, I think maybe it is time to hear about RNA and the big-bang theory from Harry Noller. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 331

Session 6: Discussions 331

Harry Noller: Sounds like we really ran out of topics.

Chris Walsh: Well, that’s the next level!

Harry Noller: Well, it starts with a question: How do you evolve the ribosome? Un- less you are a creationist, you look at this thing and you say “this is impossible”. So how to develop? If you accept the fairly widely held idea that there was something like a RNA world that preceded the protein-DNA-RNA world and which was based on the functional properties of ribosomes and other kind of functional RNAs which could then self-replicate to create new copies of themselves, then you had a genetic system that could go forward and evolve. One problem, that I’m sure everyone is distribution is allowed. distribution

aware of, is that RNA is wonderful, especially compared to DNA, in its capabilities

for folding into interesting three-dimensional structures with tertiary interactions,

9180 and so on, and evolved catalytic properties that have been shown by the ribosome people, but it is severely restricted in terms of its folding capabilities compared with only. No further No only. proteins. And also severely restricted with respect to its functional groups. So you could imagine that one early version of translation was simply to make peptides that bound to RNA and stabilized novel or only weakly-sampled conformations and there are ample examples of this. For example, the HIV Tat-TAR interaction: the vayinstitutes.be/ Tat protein binds to the TAR element of the HIV RNA to a simple bulged helix PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- with the hairpin loop at the end and induces formation of a complex tertiary interaction, a base triple, and closes the two co-axial helices to tilt by 90◦.Anyway,it

http://www.sol is a quite dramatic alteration in structure and Frankel and Puglisi and their collaborators showed a long time ago that you could mimic this with a peptide, which is

host on on host an arginine-riche peptide from the Tat protein, and then remarkably, they showed that you could even do this with arginine peptide at millimolar concentrations. So even just binding an amino acid at millimolar concentration could do this but the advantage of a peptide is that you can get away with lower concentrations. And so maybe what happened initially is that you started making peptides that greatly http://www.worldscientific.com/worldscibooks/10.1142/ enlarged the structure space of RNA and therefore its functional capabilities and its catalytic and other kind of eﬃciencies. Then, you would imagine that this would of course select for a peptidyl transferase-like activity and ancillary activity. This will then lead to larger and more interesting peptides and eventually these peptides would be sort of prosthetic groups on the RNA, contributing catalytic functions. Eventually the RNA could shrink and we ultimately end-up with the reverse, the RNA being the prosthethic group of a protein for example. An idea which people ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World think may be the remnant of an early ribosome. That is the big bang theory. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Marina Rodnina: I can continue on the big-bang a little bit. Because I think it is also interesting to see how evolution developed after the big-bang. An interesting

thing is that apparently the peptidyl-transferase center of the ribosome remains as unspeciﬁc as it was at the beginning, probably. This is the reason why Jason and Hiro are able to make their tricks with unnatural amino acids, because the ribosome October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 332

332 Proteins in Supramolecular Machines

just does not care what comes. This is probably the remainder, the fossil, if you wish, of this thing. But then with the time the evolution apparently went into the direction of speed and ﬁdelity and this led to the development of elongation factors,

which in fact only are necessary to make it all faster. The more recent evolution

apparently leads to all these proteins which are (right now we have found them as orphan proteins) often of unknown functions. And which probably one by one will be shown to do something with translation, like EF-P, EIF5A. Joachim has just told me that there is another protein which is binding to the same place, which is also one of these orphan protein which we don’t know what it is doing. It seems that the perspective is to make proteins which are better in helping the ribosome distribution is allowed. distribution than changing the ribosome itself. This is our fossil of the big-bang.

Nenad Ban: It is just a short comment on the fact that it seems that at the be- 9180 ginning there was RNA and there were lipid bilayer vesicles. It is interesting to

only. No further No only. note that today protein targeting machinery consists of an RNA element, the signal

recognition particle that maybe is due to the fact that this process is also very ancient and may have occurred at the time of the “big-bang” Harry was referring to.

vayinstitutes.be/ Jason Chin: I have two questions. One relates to how we extrapolate from small PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- number of cases to generality, I’m thinking on how Harry showed some work from Venki’s lab on the decoding center where at least, as I understand it, there is a small

http://www.sol number of structures showing coding-decoding interactions in the decoding center. More recently, for example, there was a structure of pseudouridylated tRNA with

host on on host quite diﬀerent decoding properties. We don’t, for example, as far as I know, have structure of every cognate codon-anticodon interactions to see to what extent the rules that Venki’ illustrates very nicely explain in principle those kind of structures, the genetic code wobbling, how that is a generally played out. My question is to what extent should we elucidate structurally every single case and to what extent http://www.worldscientific.com/worldscibooks/10.1142/ do we just take it on faith this is the right principle that has been uncovered? I have another question but I’ll stop.

Harry Noller: So I was careful in my talk not to say. . .

Jason Chin: It wasn’t a criticism to what you said in your talk, I was trying to understand where we should go. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Harry Noller: No, it’s a point well taken. This is a very interesting turn of events NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY to see these unusual structures from Venki’s lab and from Marat Yusupov recently that sort of change how we think about this. I would just say that the presence of

the ribosomal RNA at the scene of the crime as it were interacting intimately with these base-pairing tells us that RNA is involved intimately in this discrimination mechanism.Howitplaysoutmayturnouttobediﬀerentfromwhatwearethinking right now. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 333

Session 6: Discussions 333

Marina Rodnina: I would like to add to that. Crystal structures and structural work is not alone. There is a lot of biochemical and biophysical work from diﬀerent groups. . .

Jason Chin: Including from you?

Marina Rodnina: ...whichsomehowtestatleastpartofthesesuggestions,which come from structural models. And if they fit, it is fine. If not, something else comes. We have to reconsider that and we have to design new experiments and test it. This is a normal way in our field as we do that and I think this is one of the reasons for the progress in the last ten years. distribution is allowed. distribution

Jason Chin: Marina, do you think it’s worth getting structures of every single tRNA

9180 codon-anticodon interaction, would that be a useful activity? only. No further No only. Marina Rodnina: I would suggest to check a couple of them. From the biochemical evidence, we know which steps are the same for each type of codon-anticodon base pairing, because with biochemistry you can simply test it, it is much easier. Then you can deﬁne precise experiments where you would like to test that because

vayinstitutes.be/ biochemistry tells you there is something interesting. There I would deﬁnitively PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- make crystal structures or cryo-EM. For example, regarding the cryo-EM of the near-cognate, we have only one structure from Joachim Frank’s lab. So I would say there is still some place, some room for improvements for looking at the near http://www.sol cognate. This is still an unﬁnished story.

host on on host Harry Noller: I have a question to Nenad. Now that we have a detailed three- dimensional structure of the 80S ribosome. What is all the extra stuﬀ for? Can we generalize about that or even address a few speciﬁc topics?

Nenad Ban: Obviously, there are some hints and some of those additional functions http://www.worldscientific.com/worldscibooks/10.1142/ specifically related to eukaryotic protein synthesis, I mentioned in the talk. However, I think this is still mechanistically poorly understood and the structure hopefully will provide tools to both interpret and devise genetic or biochemical experiments to better understand these functions. That said, obviously only some new portions of the eukaryotic ribosome structures will have specific functional roles. Many of these extra features will be purely architectural. Maybe some of these features are critical during the assembly process of eukaryotic ribosomes in the nucleus. ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World What is interesting, just as an example, is that the exit site of the ribosome that NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY interacts with the membranes during insertion of membrane protein seems to be very flat and there is not that much added additional proteins or RNA mass in this

region. However, eukaryotic ribosomes have one of these RNA expansion segments that would sit right above this exit and this expansion segment has to change conformation when ribosome synthesizes membrane proteins, as it was seen in the EM reconstruction of the translocon for example in the Beckman lab. We see that October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 334

334 Proteins in Supramolecular Machines

it has to move out the way, and deletion of this helix is again lethal. So obviously it has some important role but it is not clear which. It is tempting to assume that maybe this helix could serve some sort of a regulatory role or facilitate, let’s say,

binding of certain factors that are involved during folding of proteins. And then it

moves out the way when the ribosomes have to interact with the membrane. There are many questions of that sort that will be interesting to look at in the future.

Lode Wyns: Talking about peculiar features, Joachim Frank, you showed us Try- panosome ribosomes. Could you say what is peculiar in there?

Joachim Frank: It is very intriguing. Some of these RNA expansion segments are distribution is allowed. distribution

much larger than in any other eukaryote that we know. Unfortunately, nothing is

known about these. We speculate that it has something to do with initiation and

9180 also the fact that Trypanosomes have to switch hosts. Essentially, they have to jump-start in the new host immediately in order not to perish. We see these kind only. No further No only. of expansion segments in every trypanosomal ribosome that we have looked at. So far, we cannot assign a function to these.

Rudi Glockshuber: I have a question to Nenad Ban, regarding the trigger factor.

vayinstitutes.be/ In your movie, it looked as if for a multidomain protein with say two or three PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- sequentially arranged domains, that trigger factor could only support the folding of the N-terminal domain, and once the synthesis proceeds, the trigger factor falls oﬀ but can no longer rebind. Is this true? http://www.sol

Nenad Ban: Yes, indeed. We can assume this might be the case, based on the host on on host structural features of the ribosome-bond trigger factor. But more importantly, evidence for this type of sequential folding of individual domains and multidomains proteins aided by trigger factor that cycles on and oﬀ the ribosome in a fashion that I have shown only for one domain, has been proposed and demonstrated biochemically

http://www.worldscientific.com/worldscibooks/10.1142/ with FRET studies in Ulrich Hartl’s lab and Bernd Bukau’s lab, based on the structural model that we have initially proposed.

Jason Chin: This a question for Nenad Ban. Is this phenomenon that has been reported as the translational ramping at the N-terminal region during expression. Is that related to the trigger factor?

Nenad Ban: I’m not entirely sure what you are referring to, could you elaborate a ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World little bit? NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Jason Chin: There is an increase in protein synthesis rate from the N-terminus, depending on who you talk to, over the ﬁrst 20 to 30 amino acids or something like

this, the speed of translation ramps up. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 335

Session 6: Discussions 335

Nenad Ban: Ok I wasn’t aware of that.

Jason Chin: So the question is does this somehow correlate with the trigger factor

working on the N-terminus and then being released?

Nenad Ban: Interesting idea, but I suspect that this might not be the case because the trigger factor is known to re-associate with the next segment of unfolded polypeptide. It is not present just initially.

Ian Wilson: What is really the mechanism for the diﬀerentiation between membrane proteins and soluble proteins. It seems hard to understand how you can actually distribution is allowed. distribution recognize, as it is coming out, whether this is going to fold up into an α-helix, whether it’s going to be hydrophobic enough, or is just a few hydrophobic residues? 9180 Nenad Ban: This is a very interesting question, and here we also open the question only. No further No only. of how the activities of these various factors at the exit of the ribosomal channel are spacially and temporally regulated, because you may have noticed that several of them bind virtually to identical sites at the exit of the ribosomal tunnel. Obviously, there is some sort of an interplay that is currently poorly understood. I think there

vayinstitutes.be/ is a lot of research happening in this area. To better understand for example why PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- trigger factor initially interacts with proteins and how it is displaced, or maybe for a short time, we have simultaneous binding of signal recognition particle and trigger factor in a distorted conformation, because in their energetically favored http://www.sol state they would clearly clash one with another, even though their direct binding sites do not overlap. It is a very good question and I think it is a matter of host on on host affinities. It is also clear that these factors do not have very high affinities for the ribosome. They bind with micromolar affinities, so they obviously cycle and they have some residence time on the ribosome. The bottom line is that with respect to future membrane proteins, it is not clear entirely how the signal recognition particle

http://www.worldscientific.com/worldscibooks/10.1142/ recognizes it on the ribosome. It is known, for example the Wolfgang Wintermeyer lab has reported that for a nascent chains that carry a signal sequence, already before the signal sequence exits, the ribosome increases somehow the aﬃnity for the signal recognition particle. But mechanistically, it is not clear how that could happen. In the end, obviously, there are other factors that later have to step into the game. The translocon for example, has to also bind to the same place where the signal recognition particle binds. And only recently we understand how the

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World GTP-controlled conformational change exposes simultaneously the binding site for

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY the translocon and allows signal sequence hand over from the signal recognition particle to the translocon.

Rudi Glockshuber: I have a question to all ribosome experts. I think the antibiotics which are known to bind to the ribosomes all bind to the intact ribosome, the intact quaternary structure. But there are no antibiotics that inhibit ribosome assembly. Eukaryotic ribosomes have about two hundreds assembly factors, or something like October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 336

336 Proteins in Supramolecular Machines

that and so one could easily think of assembly factors as drug targets. Why have such kind of drugs not been identiﬁed?

Harry Noller: It has been suggested by some assembly people that ribosomal assembly is so hugely cooperative that the binding aﬃnity of an antibiotic might simply

be overwhelmed by the massive cooperative interactions of all of the proteins and assembly factors.

Marina Rodnina: I have a comment on that, a very short one. The comment is that I don’t think that anybody has ever been looking for drugs against eukaryotic ribosome assembly, because I don’t know why you should do that? distribution is allowed. distribution

Rudi Glockshuber: I of course meant drugs against bacterial ribosome assembly. 9180 Marina Rodnina: The answer is very simple. This is too fast. You have to try to only. No further No only. develop a lead compound against something which takes a long time. Elongation for example takes a long time. There is a good possibility this drug would come and will bind there. The ribosomes spends most time during the elongation. Therefore the large majority of antibiotics which we have right now, are against peptidyl-

vayinstitutes.be/ transferase or against decoding (e.g., aminoglycosides). PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- Tom Muir: I want to come back to this proline eﬀect. As I understand it, at least one of the problems is that the peptidyl-tRNA ending in proline is a slightly slower

http://www.sol reactant. I have a comment that might relate to this. It turns out that peptides ending in proline, peptide esters or thioesters ending in proline, are intrinsically less

host on on host reactive as acylating agents compared to the other amino acids. It is quite a dramatic eﬀect actually. It used to be thought that was related to the conformational eﬀect with the pyrrolidine ring. It turns out that it is not the case, it rather relates to an anti-π∗ interaction involving the ester carbonyl that deactives as an acylating

http://www.worldscientific.com/worldscibooks/10.1142/ agent. That is maybe related to what you are seeing.

Marina Rodnina: Well, maybe. The only thing what we have, we can compare that to the model studies which have been done for peptide-bond formation including effects of different amino acids on the A-site substrate and P-site substrate. And from these studies, proline was not particularly different from anything else. There were more effects coming from charges. For example positive charges on the terminal amino acid led to favorable effects and a negative charge was unfavorable. Proline ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World was unremarkable under these studies and so it still remains to be seen. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Markus Aebi: I have a question to Ian Wilson. You showed this beautiful machine of these fusion proteins. How is the cell or the virus directing the synthesis of such

a loaded spring that it maintains in a high-energy state. It is synthesized in a high- energy state and only reacts when it encounters its ligand. Have you any idea on how this is done? October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 337

Session 6: Discussions 337

Ian Wilson: Yes. Clearly we are used to proteins folding in their lowest energy state, and quite clearly these fusion proteins are metastable proteins. But in a sense they are in a lower energy in as much as the environment in which they are produced.

Each of these fusion proteins in order to adopt a diﬀerent state actually has to

bind something, or to be subjected to some different chemical environment. And that varies from fusion protein to fusion protein. For influenza virus, binding on the surface of cells does not trigger any conformational changes. The virus gets taken into the endosomes and in that compartment it’s acidic. It is this acidity that actually is the trigger. Now you are in a low pH environment and that triggers the conformational change. For HIV it is a different story, you’ve got two receptors distribution is allowed. distribution and each one of them actually triggers a conformational change. When the envelope

protein binds to CD4, it actually is in a pretty closed state. When CD4 binds, the heads start to move apart and rotate around. That exposes and helps form the 9180 co-receptor binding-site, and only when CD4 binds, do you actually start to change only. No further No only. the conformation suﬃciently that the protein goes into its fusion-active form. This occurs on the membrane surface and it is not activated by pH. The second trigger is not pH, it is the binding of an additional co-receptor. So they are in a sense in a semi-stable state, they are reasonably stable proteins in a particular environment

vayinstitutes.be/ and it is only through binding ligands or change in chemical environment that you PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- come to these dramatic changes that lead to fusion.

Don Hilvert: Can I just follow up on that? Since you undergo these very large

http://www.sol conformational changes, is it possible to elicit antibodies to the intermediates using truncated constructs, to trap the fusion transition that occurs? host on on host Ian Wilson: I’m sure people have tried to look at these and to look at mutations that actually help stabilize some of these intermediates. At the moment we have not really trapped many of the intermediates because once it goes, it goes very quickly. That movie I showed you was actually a representation of the beginning http://www.worldscientific.com/worldscibooks/10.1142/ state and the end state, and you need some sort of fantasy of the in between as to how it actually goes. It’s just that we have not managed to trap an intermediate state. It may be possible for things like HIV, where you may be going into some sort of intermediate state, to actually trap that, and you in fact can trap this stage with antibodies.

Kurt W¨uthrich: Is there some protein-related new chemistry in these macromolecu- ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World lar assemblies? What I’m looking for is perhaps best illustrated with work that Don

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Hilvert has been involved with some time ago, work on enzymatically active antibodies, which have not been discussed at all here. What impressed me at the time, and actually I don’t quite understand why this is not carried on and emphasized

more clearly, is that organic synthetic reactions that would otherwise only happen in extremely dry non-aqueous solvents, run in aqueous media in these enzymatically active antibodies. Do you have something similar to report about chemistry October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 338

338 Proteins in Supramolecular Machines

in these macromolecular assemblies which would be unlikely to happen outside of these “machines”?

Judith Klinman: I just want to say many enzyme active sites have very low dielectric constants, in which case you don’t really need a machine. You just need to control

the local dielectric constant to drive reactions that normally would go in an organic solvent.

Kurt W¨uthrich: Is it not surprising to you that the Diels-Alder reaction would run in an enzymatically active antibody in aqueous solution? distribution is allowed. distribution Judith Klinman: Well, it is a surprising reaction. I just think the degree of hydration and the way in which the active site residues control the dielectric, the proximity

9180 for sure is a factor, but no I don’t ﬁnd it that astonishing. only. No further No only. Marina Rodnina: Perhaps I want to answer this question in a little bit diﬀerent way. We are dealing with normal translation, but we know that people are introducing amino acids and making products which have not been there before. This is more a question to Jason [Chin] and to Hiro [Suga], who make all these fantastic products.

vayinstitutes.be/ This is all was ribosome does. It just can take a number of substrates and unnatural PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- substrates which have never been there before.

Don Hilvert: There are still amide bonds. http://www.sol Nenad Ban: Maybe I can just ask Craig Townsend to comment a little bit on the

host on on host strategies to develop potentially new natural products with polyketide synthases, are there “surgical” procedures to redesign the order of their active sites or to design new types of multi-enzymes.

Craig Townsend: Right, well, for the last twenty years people have had a great deal http://www.worldscientific.com/worldscibooks/10.1142/ of interest in doing this with modular systems. I think the diﬃculty there is that as you proceed through each individual domain that has been optimized evolutionary, it is hard to get a good overall yield. Our experience is limited, I did not have time to talk about this, but it is possible to make hetero-combinations of these individual catalytic domains that have a similar function and the remarkable thing is that they will work together. They will also carry out synthesis in a way that — we call programming because we don’t understand exactly what controls these events — if ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World a particular ring system is made by a PT-domain for example, it can be supplied

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY with a polyketide precursor that is shorter or longer, and amazingly it still binds and does the cyclization chemistry faithfully. So it would be wonderful to have structural information of these individual pieces to have some idea of what controls

these events, but I think the intrinsic flexibility of binding can be taken advantage off. I am sure it can be optimized, there is doubtlessly downstream evolutionary kind of things that one would want to do to make them efficient, but the intrinsic activity is quite good. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 339

Session 6: Discussions 339

Jason Chin: I have a question for Craig Townsend and maybe for Nenad Ban about assembly line biosynthesis. Where are we with understanding at the mechanistic level, by which I don’t mean chemical mechanism, I mean structure and dynamics of

the protein and how that relates to catalytic steps or transfer steps in term of both

dynamics and structure. Is that the missing information for moving this forward or is there something else that would be more important? What is your view?

Craig Townsend: This is a complicated question because the intrinsic fact for those iterative and modular systems is that the intermediates are all bound, there are no free intermediates, so a lot of the standard kinds of biochemical tests that you distribution is allowed. distribution

would like to be able to apply aren’t so easy to do. There are a lot of reactions, if

you count up, just for making a fatty acid where things are very nicely orchestrated to be repeated every single cycle, it’s ﬁfty reactions that take place. The individual 9180 steps, rates and ﬂux and so forth, people have worked at this, but it’s complicated. only. No further No only. Jason Chin: Are there biochemical ways to trap these intermediates?

Nenad Ban: I would like to comment on one other aspect. Currently, the structural information on the fungal, mammalian type of fatty acid synthases gives us

vayinstitutes.be/ a relatively good idea of the basic architecture of a single module of a polyketide PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- synthase. However, we have relatively limited information on how these modules assemble into complete polyketide synthases that have to shuttle substrates from one module to another for each step of the series of reactions. I think this is deﬁnitively http://www.sol a very interesting future direction of research in understanding both the structure and the dynamics of these machines. host on on host

Wilfred van der Donk: I saw David Sherman give a talk. In that case it was I think a PKS or NRPS, I don’t remember. They had done cryo-EM. It was really quite diﬀerent from the fatty acid synthase. Even the way the individual domains were

http://www.worldscientific.com/worldscibooks/10.1142/ organized, I was surprised. Instead of being out, in kind of like a butterﬂy type, they moved back in and they looked very diﬀerent. I think there are still a lot of things to be understood there.

Chris Walsh: This polyketide synthase, would there be a simple trick that it can make a complex polymer or it could be another processing enzyme? There is a very big interest in diﬀerent materials with diﬀerent properties as well, but often

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World to make them solid actually, it is good to have them in a polymer state. Here you will be able to have kind of a polymer with a certain sequence of chemistry. NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY

Craig Townsend: That is possibly true. There is a limitation in the iterative systems

in that there are essentially four ring types that are made, right now. You can make fused rings, up to six for sure, maybe four fused together. Then you will be at the end of what is currently possible to do in a product template domain. I think it will be diﬃcult to make an extended condensed polymer. Maybe pieces linked together individually, groups of three and four perhaps, that might be possible to do. October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 340

340 Proteins in Supramolecular Machines

The other problem is the intrinsic reactivity of the poly-β-ketone. It is remarkable, it is not something that you would be easily able to do in a laboratory, to make something like that. The trick is a kinetic one, one thing that I would have liked

to have said but did not say, is that the populations, in an iterative system, of the

individual active sites are essentially zero. Except for the malonyl transferase. So the syntheses that initiate chain extension are very rapid. You can actually make a poly β-ketone suﬃciently quickly. That is, it hasn’t had a chance to actually react with itself, and that’s then controlled in a template domain by being bound and cyclized in a fairly rigid and deﬁned way. So it takes quite a bit of engineering to change that beyond about four rings. distribution is allowed. distribution

Don Hilvert: I might add in that context, that there are enzymes that polymerize activated acids to make poly-esters, poly-hydroxybutyrate for example. You can 9180 imagine adapting such systems or copying that strategy to make artiﬁcial poly-

only. No further No only. merases.

Nenad Ban: I have one more question to Rudi (Glockshuber). It is interesting to see that these pathogenic bacteria attach to the cell walls using the pili. It is curious to think of what sort of forces would be necessary to detach the pili. Has vayinstitutes.be/ anybody done maybe some, let’s say, optical tweezers experiments to see whether PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- the subunits will unfold or whether they will break or whether they will pop out the anchor site on the bacteria. What is the weakest point? http://www.sol Rudi Glockshuber: The first observation is that you have this apparent increase in binding affinity to the receptor upon application of sheer force. The hypothesis is host on on host that this is related to the domain separation in the adhesin and a conformational change in the lectin domain that then has a lower off rate for the ligand. If you continue pulling, then you see the unwinding of the helical pilus rod. And a gain in length by about a factor of two. And only then, they fall apart. http://www.worldscientific.com/worldscibooks/10.1142/ Nend Ban: Presumably then unfolding?

Rudi Glockshuber: Or by breaking donor strand domain interactions, that is not entirely clear.

Don Hilvert: I also have a question to Rudi (Glockshuber). How sensitive is this F-tag to mutations and can you use the tag itself as an inhibitor for pilus formation? ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY Rudi Glockshuber: This is a very good question. We have not tested the peptide for inhibitory activity. One could add it to the periplasm, by just adding it to the medium and it would maybe diﬀuse to the periplasm. What we have done:

the FimG-dsF peptide complex is the most stable non-covalent protein complex known to date. But it has actually quite a bad on-rate of only 300 M−1sec−1. We have tried to increase charge complementarity between peptide and protein by October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 341

Session 6: Discussions 341

introducing mutations in the protein next to the peptide binding-site and also by mutating the peptide residues which point towards the surface. We have now, it is not published yet, a new combination of FimG variant/DsF variant which has

10−22 molar dissociation constant, and an about one hundred times better on-rate.

Don Hilvert: That could be very useful then for competitive inhibition with the natural system.

Rudi Glockshuber: Yes, but we have not tested it yet.

Chris Walsh: I would like to go to another question. It is about electron microscopy, distribution is allowed. distribution but also about the pili. So in 3D electron microscopy, we can do this with the ribosome because it is very big. We hope that we will get smaller but we will have a

9180 problem to get below 200,000. The question I have is, can we use these pili ﬁlaments to insert relatively small molecules? And then have a helical architecture that we only. No further No only. canlookatinEM,whichisnicelypresented as a helical structure. Is there any insertion point we could use in the pili?

Rudi Glockshuber: I believe that it is diﬃcult. Because the translocation pore of

vayinstitutes.be/ the assembly platform is just wide enough to allow translocation of an individual PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- folded subunit. One can also reconstitute the pili after dissociation and let them reassemble. The pili become a little bit shorter, or signiﬁcantly shorter, but still ﬁfty nanometers or so long. This is maybe possible but very likely not with the http://www.sol assembly platform.

host on on host Chris Walsh: Joachim, how far are people with using scaﬀolds for electron microscopy?

Joachim Frank: I am not aware of any use right now. Long time ago, we used the ribosome itself in order to ﬁnd the position of an rRNA by inserting a tRNA http://www.worldscientific.com/worldscibooks/10.1142/ genetically, so we found the tRNA stuck at the outside. I think people have been trying to do similar experiments. One with the group II intron. We have a ﬁrst reconstruction of a group II intron in collaboration with Marlene Belfort. But we need to identify the components since the reconstruction is at low resolution. One idea is to make a similar kind of insertion.

Rudi Glockshuber: We still have about ten minutes. There are so many exciting ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World developments in electron microscopy, we also have free electron laser technology

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY and so many experts are here. I would like to propose that we maybe spend the next ten minutes on the future of structural biology. What are your thoughts where structural biology, in particular of supra-molecular structures, will stand in ten or

twenty years from now? Do you agree with this?

Bernard Henrissat: I would agree if I can ask a last question before? This is on the pili. I would like to know what is the conservation of this machinery across bacterial October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 342

342 Proteins in Supramolecular Machines

species? Do we expect the same properties always or is there diversity there? Second, are there any interactions between the pili and other surface components of the cell such as peptidoglycans?

Rudi Glockshuber: With respect to peptidoglycan, I do not know any reports.

Similarity: the pair-wise sequence identity within type 1 pilus subunit is not more than 35% or so. Other pili from Gram-negative bacteria which are also assembled according to the same mechanism share sometimes even less than 20% sequence identity with the corresponding subunit from other species. But structurally they are all identical. There have been a couple of experiments, or reports where it was distribution is allowed. distribution

tried to replace for example the chaperone from one pilus system from a certain

species by the chaperone from another species. Sometimes it worked, sometimes it did not. Overall, I think the general mechanism that I described is the same for 9180 this kind of pili. only. No further No only. Jason Chin: I have a question. We don’t really have a representative of the other side of the argument here, but I guess I’m curious what the panel think about the relative merits of optical methods, of subdiﬀraction imaging, so-called super- resolution, and where that’s going to go and how that is or is not useful and might vayinstitutes.be/ complement electron microscopy? I guess Joachim [Frank] might be the best person PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- to start.

Joachim Frank: I really don’t know the range that is still possible in the light http://www.sol microscopy ﬁeld, but I’d like to mention something that has not been mentioned before, which is the power of correlative methods, where LM is combined with cryo- host on on host EM. It is possible now, through the development of certain stages, to follow a live process in detail and then freeze it at a particular moment. And then keep track of the coordinates and transfer everything into the electron microscope. Even now a stage is being developed, in which everything stays in the electron microscope http://www.worldscientific.com/worldscibooks/10.1142/ for the duration of both LM and EM imaging. This is a very exciting future of an approach in which one can study something that is happening in vivo and then immediately take a look at a much higher magniﬁcation.

Harry Noller: I also had a question, probably for Joachim [Frank]. Can one combine electron tomography with the direct electron detection technology to now solve the cell and beginning of course with the shallow? ©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Joachim Frank: Yes, absolutely. The ﬁrst results that we are obtaining show that NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY we have a much better resolution in tomography. It is very exciting, but it is also very humbling because we are now dealing with incredibly large volumes. We have

to sort of disentangle these. The computational eﬀorts are really mind boggling there. One of the nice things with these direct detection cameras is that they allow you to do multiple frames. So, the K2 summit camera from Gatan catches forty diﬀerent frames per second. In this case, any motions of the specimen that might October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 343

Session 6: Discussions 343

be beam-induced or are due to instability of the instrument can be compensated. This is part of the reason for the improvement in resolution. So absolutely, electron tomography is one of the big winners in this game.

Gebhard Schertler: So we had a huge development in X-ray diﬀraction, what a

free electron laser can do is give you tremendous intensities, so a huge number of photons in a very short time. Here you see, we have femtoseconds up there. What this diagram here shows, is actually a very important simulation done in 2000, where actually it was proposed that every atom would maybe absorb an X-ray photon that is not the diﬀracting photon but then electrons leave the molecule and distribution is allowed. distribution

actually there is a Coulomb explosion from the residual there. How can that be

useful? If you make the X-ray pulse short enough, like 2 femtoseconds in this case, we can actually see the molecule before it explodes. 9180 Here you see the second important thing how this was practically made useful.

only. No further No only. We are injecting now little crystals or particles, or we could do ribosomes, into

the machine with a liquid jet and the little light you see there is actually the destructive power of the free electron laser but we can record these patterns, and now, at room temperature, structures have been solved to 1.5 A˚ resolution of various things, new molecules and also very recently of membrane proteins with a slightly vayinstitutes.be/

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS modiﬁed technique. So we really look at a very new technique where we can also

- do spectroscopy and other methods in completely diﬀerent timeframes. Another important thing about this technique is that we have now laser like X-ray light,

http://www.sol so holographic techniques, or phase reconstruction and retrieval will really change dramatically over time, so I think we have a fantastic development but actually it is

host on on host working already, it’s not that we need years to develop, we can apply it to biological molecules now.

Rudi Glockshuber: Maybe one last comment and then we have to ﬁnish.

http://www.worldscientific.com/worldscibooks/10.1142/ Ian Wilson: So looking at fantasy, but may be it isn’t so much fantasy, I mean doing single molecules. That’s the ultimate, you don’t have to get a crystal anymore and you can do everything oﬀ a single molecule. Whether that’s actually going to pan out, at least that’s the way people are thinking now, that would really revolutionize what we can do, if we can get high resolution information from single molecules.

Gebhard Schertler: But there is a big debate whether EM isn’t better at it, so I think

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World we have to ﬁnd the right niche for every method. I am not convinced that actually

NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY free electron lasers in the first generation will be very good for single particles, they are very good actually for difficult membrane proteins, difficult crystallization projects, in giving you high resolution but at the moment they are not a single molecule method.

Joachim Frank: To comment on this, as I pointed out when I spoke, we are really on the verge of atomic-level resolution for several big enough macromolecules and October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 344

344 Proteins in Supramolecular Machines

we will see this reﬂected in the literature in the next two or three years; we will soon have an entire promenade of structures captured in actual ‘live’ situations; in other words macromolecular machines ‘caught in the act’ and so this will be a very

exciting prospect in future research.

Gebhard Schertler: I think the other exciting thing is really the combination and spread, what we need is methodology that spans the whole timeframe, the whole spatial frame, and one single technique most likely cannot do that.

Rudi Glockshuber: Ok, then this is I think quite a good ﬁnal statement on this part of the discussion. I would like to thank all the speakers of this morning session, distribution is allowed. distribution

and those of you who contributed to the discussion for their excellent contributions,

thank you very much. 9180 only. No further No only.

vayinstitutes.be/ PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 345

345

Index

α-lactalbumin, 156 streptomycin, 103 α-synuclein, 261 thiocillin, 3 distribution is allowed. distribution β-methylene, 96 vancomycin, 237

β2microglobulin, 261 antibodies, 45, 102, 107, 110, 127, 138, 16S RNA, 130, 147, 162, 308, 310, 311, 139, 158, 159, 167–171, 179, 186, 192, 9180 327 193, 221, 222, 230–235, 243–246, 18S rRNA, 320 249–252, 254–260, 262, 263, 266–284, only. No further No only. 80S ribosome, 286, 333 293–296, 324, 325, 337, 338 antibiome, 276 activity-based protein probes (ABPPs), anti-PD1, 282 131–133 anti-TLR4, 282 adenosine, 216 anti-TNF-α, 272 vayinstitutes.be/ adhesin, 287, 288, 326, 340 binding, 102 PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- ADP-ribosylation, 21 bispeciﬁc, 282 adrenaline, 102 carbohydrate-binding, 127, 169 agonist, 190, 192, 193, 196, 212, 214, catalytic, 61, 84, 106 http://www.sol 215, 220–222, 257, 266, 270, 281, 283 cell-mediated cytotoxicity, 159 agonist antibodies, 211, 257 conformational, 107

host on on host AKT2 inhibitor, 275 drug conjugate, 232, 255, 269, albumin, 84, 223 277–279 allostery germline, 171 allosteric eﬀect, 104 glycan interaction, 170 allosteric modulator, 107, 189, 193, hypermutation, 260

http://www.worldscientific.com/worldscibooks/10.1142/ 219 IgG, 234, 254, 270, 284 amber stop codon, 79, 80, 109 immunocytokines, 255 amber suppression, 109 immunogens, 165, 171, 296 amiloride, 214 intrabodies, 258, 269 aminoacylation, 79, 98, 239 monoclonal, 139, 231–233, 243, 254, aminoacyl-transferase, 239, 301 271, 284 aminoacyl-tRNA, 305, 308 multivalency, 39, 127, 168 aminoacyl-tRNA synthetase, 80, 81, 98, nanobody, 192–194, 207, 234, 246,

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 109 260–263, 267, 278 amylin, 44 PEGylation, 244 NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY anabolism, 169 phage display, 250 anti-amyloid β-anticalins, 269 protein interaction, 170 antibiotics, 3, 4, 6, 37, 84, 91, 103, 130, anticalin, 234, 243–245, 269

134, 163, 237, 290, 291, 299, 320, 321, artiﬁcial enzyme, 39, 59–62 325, 335, 336 attendees daptomycin, 103, 237 Aebi, Markus, 158, 160, 164–167, erythromycin, 103, 299 169, 174, 175 October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 346

346 Index

Arnold, Frances H., 102, 103, 107, Steyaert, Jan, 107 109, 110, 112, 113 Suga, Hiroaki, 116 Ban, Nenad, 318, 323, 325, 326, Townsend, Craig A., 174 328, 329, 332–335, 338–340

van der Donk, Wilfred A., 32, 33, Chin, Jason W., 40, 43, 46, 47, 54, 35, 38, 39, 42–44, 46, 47, 52, 53,

96–98, 108, 109, 166, 167 108 Edwards, Aled M., 31, 32, 34, 35, von Heijne, Gunnar, 179, 203, 209, 38, 55, 169 210, 214, 224–226 Frank, Joachim, 324, 326, 330, 334, Walsh, Christopher T., xxii, 31, 341–343 37–44, 46, 51, 52, 54, 55, 103, Gerlt, John A., 37, 95, 98, 99, 105, 104, 108, 277–279, 281, 282, 330, 108, 114, 157, 158 331, 339, 341 distribution is allowed. distribution Glockshuber, Rudolf, 52 Wilson, Ian A., 51, 154–157, 159,

Godzik, Adam, 31–36, 43, 49, 163, 165, 167–170, 172–175 51–53, 154–156, 162, 163, 174 Wolan, Dennis W., 154–156, 161, 9180 Gr¨utter, Markus G., 162, 165 162, 174 Henrissat, Bernard, 51, 110, Wong, Chi-Huey, 111, 158–160, only. No further No only. 155–157, 162, 172, 175 163–165, 168–170, 173, 175 Hilvert, Donald, 45, 53, 54, 94, W¨uthrich, Kurt, 41, 42, 96, 97, 108, 96–98, 102, 103, 108, 111, 112, 112, 163, 171, 174 116, 172, 173 Wyns, Lode, 47, 110

vayinstitutes.be/ Houk, Kendall N., 98, 100, 106, auditors

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS 112, 113, 115 Amory, A., xviii

- Klinman, Judith P., 48, 49, 55, Bartik, K., xviii, 31 94–96, 99, 100, 104–106, Bruylants, G., xviii 111–115, 155, 164 Callewaert, N., xviii, 154, 175 http://www.sol Lekkerkerker, Henk, 109, 110 Collet, J.-F., xviii, 323 Lerner, Richard, 44, 45, 49, 52, De Wit, A., xviii host on on host 102–104, 107, 155, 159–161, 170, Deroo, S., xviii 171 Fr`ere, J.-M., xviii, 94 Locher, Kaspar, 151, 208–210, Goormaghtigh, E., xviii, 31 222–225, 276 Govaerts, C., xviii, 203 Muir, Tom W., 33, 37, 38, 40–43, Homble, F., xviii

http://www.worldscientific.com/worldscibooks/10.1142/ 46, 47, 54, 107, 161, 164 Lafontaine, D., xviii Neri, Dario, 159 Lesage, A., xviii, 266 Noller, Harry, 326, 327, 330–333, Loris, R., xviii, 154 336, 342 Marion, P., xviii Orengo, Christine, 31, 32, 35, 37, Muyldermans, S., xviii, 266 43, 47–53, 162, 172 Parmentier, M., xviii, 203 Reymond, Jean-Louis, 32, 39, 40 Raussens, V., xviii Rodnina, Marina, 106 Remaut, H., xviii, 323

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Roth, Brian, 45, 103, 104, 161, 162 Ruysschaert, J.-M., xviii Schertler, Gebhard, 40, 41, 50, 101, Vandenbranden, M., xviii NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 114, 158 Vers´ees, W., xviii, 94 Seeberger, Peter H., 160, 163, 164, 167, 168, 171–173, 175 BACE2, 261

Sidhu, Sachdev, 266, 267, 272, 275, bacteria, 16, 23, 77, 103, 120, 122, 130, 276, 279, 281, 283, 329 131, 136, 139, 142, 143, 147, 155, 157, Skerra, Arne, 39 158, 161–163, 174, 175, 180, 199, 225, Stevens, Raymond, 106 261, 291, 305, 315, 319, 320, 328, 329 October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 347

Index 347

Actinomyces,34 grazing incidence X-ray diﬀraction, aerobe, 3, 34, 35, 54, 55 138 anaerobe, 3, 34, 35, 54, 55, 87, 121 intravital ﬂuorescence microscopy, Bacillus anthracis, 139 269

Bacillus megaterium,85 NMR spectroscopy, 13, 97, 182, 183

Bacillus subtilis,44 X-ray crystallography Bacteroides thetaiotaomicron, 118, molecular replacement, 179, 324 121, 122, 162 X-ray structure, 67, 76, 96, 97, 112, Campylobacter jejuni, 151 179, 192, 199, 200, 241, 245, 275, Clostridium, 155 276, 290, 295, 301, 313, 324, 326, Clostridium diﬃcile, 138 329, 330, 343 Escherichia coli, 47, 80, 81, 87, 97, biophysical tools, 62, 106 distribution is allowed. distribution 98, 151, 165, 179, 234, 243, 246, biosensor, 46, 264

287, 288, 290, 301, 306 biotic, 54 Gram-negative, 342 blood-brain barrier, 223, 269 9180 Lactobacillus, 157 bond rotation, 94 microaerophilic bacteria, 55 brown fat, 259, 281, 284 only. No further No only.

microbes, 23, 25, 36, 38, 54, 123, butyrate, 148, 154, 155 142, 146–148, 157 Paracoccus denitriﬁcans,77 Caenorhabditis elegans, 80–82 pathogenic bacteria, 130, 134, 147, calcitonin, 42

vayinstitutes.be/ 199, 340 carbohydrate, xxi, 26, 39, 40, 117, 118,

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS Pelagibaca bermudensis,76 122, 125, 127, 133, 136–139, 141–144,

- pili, 287–289, 325, 326, 340–342 146, 147, 149, 150, 152, 154, 156–158, pilin domain, 122, 288, 291 160, 163, 165, 166, 168, 170, 172, 174, pilus, 122, 287–290, 325, 340, 342 175, 273, 291 http://www.sol pilus assembly, 122, 289, 290, 325 carbohydrate-based vaccines, 127 Prochlorococcus,8 carbohydrate-processing, 121 host on on host prokaryote, 3, 34, 91, 319, 330 catabolite, 99 Sodalis glossinidius, 264 CD4, 294, 337 Streptococcus pneumonia, 138, 139, cells 163 293T, 325 Streptomyces, 44, 47, 103 B-cell, 170

http://www.worldscientific.com/worldscibooks/10.1142/ thermophile, 111 CD34 positive, 259, 282 binding energies, 99 dendritic, 161, 170, 259 biodistribution, 247, 268, 284 insect cell, 166, 169 bio-ethanol, 109, 110 stem cell, 120, 127, 211, 257, 266, biophysical techniques 282, 283 electron crystallography, 179, 180, T-cells, 161, 170 313 chaperone, 151, 193, 225, 260, 261, electron microscopy, 180, 295, 313, 288–290, 342

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 314, 317, 330, 341 chemical ecology, 33 Gatan, Inc., 342 chemical probe, 29, 131, 156 NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY TM K2 summit direct detection chemical reactions camera, 342 1,3-dipolar cycloaddition, 239 electron tomography, 314, 316, 330, acetylation, 21, 37, 81

342 acylating agent, 336 free electron laser technology, 341, acylation, 38, 46 343 amidation, 44, 45 FRET, 309, 314, 316, 334 Birch type reduction, 35 October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 348

348 Index

bromination, 54 computation C-H activation, 96, 105, 106, 113 bioinformatics, 4, 37, 43, 45, 51, 70, chlorination, 46, 54 90, 156, 163, 301 “click” chemistry, 132, 133 DARPA, 62, 114 cycloaddition, 61, 80 docking, 76, 98, 99, 106, 108, 188, cyclopropanation, 85–88 219 dehydrative aromatization, 4 enzyme design, 70, 71, 84, 114 Diels-Alder, 60, 61, 72, 82, 338 informatics, 33, 46, 71 enantioselectivity, 85–87, 107 ligand docking, 75, 78, 189 ﬂuorination, 54 metaHIT, 24 GTP hydrolysis, 105, 213, 311, 312 prediction, 112 halogenation, 54

distribution is allowed. distribution QM/MM approach, 68, 99

H-D exchange, 96 quantum mechanics, 66, 68, 70, 71, hydroxylation, 45, 85 73, 100, 106, 107 Kemp eliminase, 61, 72, 97

9180 Rosetta, 70, 72 methyl transfer, 65, 67, 68, 94, 105, SABER, 72 111, 113 only. No further No only.

screening, 112 methylation, 21, 28, 37, 38, 81, 91 virtual ligand screening, 189, 218 mixed anhydride, 173 Consortium of Functional Glycomics, regioselectivity, 107 167 chemo-enzymatic approaches, 173 corrinoid, 34

vayinstitutes.be/ chemoinformatic methods, 99 co-translational, 166, 224, 329

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS chemokine, 138, 189, 216, 220

- cresol, 154 chimera, 311 cresol sulfate, 154 cholesterol, 183, 186, 189, 190, 205–209, CRISPR, 169 223 http://www.sol CXCR4, 189, 216, 294 cholesteryl hemisuccinate tris salt, 207, cytokine, 138, 231, 233, 246, 254, 255, 208

host on on host 257, 268 chromatin, 20, 21, 28, 37, 41, 46, 81 classiﬁcation D-amino acids, 76, 239, 251, 280 CATH superfamily, 18 DARPins, 234, 273 CAZyme, 172 dediﬀerentiation, 281 FunFHMMer, 17

http://www.worldscientific.com/worldscibooks/10.1142/ gene anthology, 6, 34 deglycosylate, 167 gene ontology, 51, 53 dendrimer, 11–14, 39, 40 GeneBank, 23 deuterium, 67, 96 KEGG database, 99 dielectric constants, 338 Pfam database, 24, 25, 148 digestive tract, 144, 155 phylogenetic tree, 17, 49, 50, 178, directed evolution, 11, 59, 61, 70, 72, 187, 220 84–86, 96, 102, 114, 303 protein families, 24, 25, 28, 29, 34, diseases

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 119, 147, 148, 186, 243 amyloidosis, 261, 262 UniProt sequence, 17, 23, 24 ankylosing spondylitis, 272 NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY UniProtKB/TrEMBL, 75 basal cell carcinoma, 213 clozapine, 219 cachexia, 281 cMET binders, 275 colorectal cancer, 130, 154, 155

colon, 155 Creutzfeldt-Jacob disease, 171 combinatorial, 250, 257 diabetes, 24, 122, 130, 261 combinatorial libraries, 103, 193 diabetes insipitus, 212 combinatorial method, 281, 282 endometriosis, 255 October 14, 2014 16:39 WSPC Proceedings - 9.75in x 6.5in master/pg 349

Index 349

inﬂammatory bowel disease, 120, discovery, 154, 189, 195, 197, 203, 130, 134, 148, 255, 272 206, 209–211, 216, 235, 242 insulinoma, 281 erythropoietin, 159, 160, 164, 231, 244, 266

leukemia, 282 lymphoma, 280, 282 GDC0449, 213 melanoma, 282 generics, 271 microbiome-associated diseases, 131 high-throughput assay, 156 obesity, 130, 217 immunosuppressive, 161 psoriasis, 159, 255, 272 lovastatin, 72 psoriatic arthritis, 272 maytansine, 279 rheumatoid arthritis, 234, 255, 272 penicillin, 281, 299 Plavix, 211

distribution is allowed. distribution schizophrenia, 206

Shwachman-Bodian-Diamond structure-based design, 219 syndrome, 321 Travina, 206

9180 trypanosomiasis, 260, 262, 263 ulcerative colitis, 130, 148 ecologic communities, 34 ecologist, 33 only. No further No only.

domains of unknown function, 25, 31, 108 electron microscopy, 316, 318, 342 de-orphanize, 35, 108 cryo-EM, 261, 286, 308, 309, 311, orphan, 36, 104, 195, 206, 219, 332 315, 316, 324, 326, 333, 339 cryogen, 326 orphan families, 35

vayinstitutes.be/ focused ion beam, 314, 330 unannotated protein, 55

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS electrophilic, 54

- dopamine, 67, 68, 196, 216 electrostatics, 17, 48, 59, 105, 106, 189 Drosophila melanogaster, 80–82 elongation factor, 332 drugs EF-P, 305, 306, 328, 329, 332 http://www.sol Ambrex, 278 EF-P experiment, 328 antagonists, 196, 212, 214, 220, 221, endosome, 212, 264, 294, 337

host on on host 266, 281, 290 enthalpy, 105 antibacterials, 103 entropy, 105 antibiotics, see also antibiotics enzymatic activities, 45, 75, 76, 131, antibodies as drugs 132, 308 Enbrel, 272 aldolase, 12, 60, 157

http://www.worldscientific.com/worldscibooks/10.1142/ Herceptin, 232, 233, 276, 284 aminoglycoside, 336 HUMIRA, 271, 272 ATPase, 223 Mylotarg, 279 barrier compaction, 65 Prevnar, 163 binding constant, 95 Raptiva, 159 carbohydrate-active enzymes, Remicade, 272 141–144 Retuxan, 272 carboxylase, 53 Trastuzumab, 232, 284 catabolism, 77, 157

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World anti-inﬂammatories, 103, 159 catalyse, 39, 53, 116 antimicrobial, 7, 43, 44, 306 catalysis, xxi, 4, 12, 26, 48, 59–62, NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY biosimilar, xxi, xxii, 159, 160, 229, 64, 65, 67, 68, 70–72, 75, 77, 79, 231, 233, 235, 237, 239, 241, 243, 80, 82, 84, 88, 90, 92, 94, 95, 99, 244, 247, 249, 250, 252, 254–256, 100, 104–107, 111, 115, 116, 122,

266, 270, 272–274 289, 300, 303, 308, 309, 328, 329 chemotherapy, 282 catalytic eﬃciency, 62, 106 CPT-11, 174 catalytic triad, 48, 72 cyclosporine, 81, 237 catechol methyltransferase, 100 October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 350

350 Index

cellulase, 109, 142 transporter, xxi, 77, 177, 179, 180, cellulose, 109, 110, 141 192, 201, 203, 204, 210, 214, C-methyltransferase, 4 222–224, 306 deacetylase, 241 tunneling, 65–68, 95, 96, 105, 106,

dehydrogenase, 77, 157 111

demethylase, 29 enzymatic f activities dioxygenase, 76 hydrolase, 118 dwell time, 210, 211, 266 Enzyme Function Initiative, 34, 75, 108, electrostatic catalysis, 64 114 enzymology, 39, 59, 75, 81, 111 ER lumen, 166 epimerase, 76 eukaryote, 3, 23, 34, 36, 49, 91, 131, esterases, 12, 71, 142 150, 151, 199, 225, 306, 315, 319–321, distribution is allowed. distribution fatty acid synthase, 300–302, 339 329, 330, 334

ﬂippases, 201 eukaryotic translation initiation factor ﬂuorinase, 54 5A-1, 332 9180 fucosyltransferase, 127, 169 evolutionary analysis, 49

only. No further No only. glucuronidase, 174

glycosyl transferase, 125, 150, 152, Fc receptor, 159 156, 157, 165, 169, 172, 175 ﬁbrillogenesis, 262, 283 hydrolase, 23, 121, 152, 157 FimG-dsF, 340 kinase, 23, 31, 34, 38, 41, 45, 46, ﬂexible in-vitro translation, 239

vayinstitutes.be/ 82, 106, 110, 157, 187, 189, 204, Flexizymes, 238, 239

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS 218, 220, 241 ﬂora, 134, 158

- kinetic, 67, 165, 211, 289, 305, 340 ﬂuorescent protein, 104, 105, 270 lysine amino mutase, 92 ﬂuorine label, 212 lysozyme, 156, 261–263 fragment-based drug discovery, 218 http://www.sol malonyl transferase, 340 FunTree, 17 mannosidase, 174 fusidic acid, 309, 311 host on on host methyltransferase, 29, 31, 67, 95 fusion protein, 207, 244, 255, 336, 337 oxidase, 76, 289 oxidoreductase, 151 gene clusters, 12, 162 P450, 4, 54, 58, 85–88 genome mining, 6–9, 33, 47, 59, 109 peptidyl-transferase, 328, 331, 336 genome sequence, 6, 28, 29, 75, 119,

http://www.worldscientific.com/worldscibooks/10.1142/ PNGase F, 225, 325 142, 143 protease, 60, 142, 170, 189, 237, genomics, 16, 23, 54, 131, 137, 156, 179, 238, 251, 252, 261, 294 249, 252 proteasome, 167 glycobiology proteolysis, 44, 252 deglycosylation, 166 racemase, 60, 99 ER glycosylation, 169 retroaldolase, 97 GCSF, 266 ribonucleotide reductases, 90, 91 glucagon, 42, 183

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World RNA polymerase, 41, 164, 305 glycan, 122, 125–127, 136–139, 141, RNAse, 142, 164 151, 152, 158, 159, 163–165, NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY sialyl transferase, 169 167–171, 173, 175, 233, 290, 295, thermolysin, 173 324, 325 thioesterase, 301 glycocalyx, 164

transferase, 306, 308, 328, 331 glycocode, 152 transition state, 59, 64, 67, 68, 71, glycoform, 126, 159, 166, 270 94, 99, 100, 106, 107 glycolipid, 138, 210 transition state theory, 64, 115 glycome, 163, 167 October 23, 2014 9:30 WSPC Proceedings - 9.75in x 6.5in master/pg 351

Index 351

glycoprotein, 39, 118, 125, 126, 136, Grubbs–Hoveyda type catalyst, 61 141, 152, 158, 160, 164, 165, 169, GTPγS, 214 173, 175, 233, 287, 293–295 glycosaminoglycan, 136–139, 141 hemagglutinin, 126, 158, 159, 170, 293,

glycoscience, 125, 127 295

glycosidic bond, 170 heparin, 138, 167 glycosylation, 126, 127, 139, high-throughput structural biology 150–152, 158, 159, 163–167, (HTSB), 119 171–173, 225, 255, 271, 272, 294, histone, 20–22, 46 325, 329 modiﬁcation, 20, 37 glycosyl-glycan, 167 tail, 38 high mannose, 166, 175, 324, 325 HIV, 43, 60, 81, 165, 168, 171, 174, 175, distribution is allowed. distribution high mannose patch, 174 324, 331, 337

mannose, 287, 325 gp120, 169, 174, 294, 295 monosaccharide, 126, 141, 172 gp140 Env trimer structure, 294, 9180 mucosal layer, 155 295, 323 N-glycosidic linkage, 165 HIV-1, 293–296 only. No further No only. N-glycosylation, 151, 160 treatment, 281 oligosaccharide, 39, 125, 127, vaccine, 165 136–138, 150–152 holographic technique, 343 trimannose, 160 homeostasis, 92, 130, 134, 161, 245

vayinstitutes.be/ unfucosylated, 271 homologue, 16, 17, 32, 35, 36, 40, 48

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS variant surface glycoprotein, 262, horizontal transfer, 162

- 267 human gut, xxi, 19, 25, 43, 119–121, Golgi, 151, 152, 165, 166, 169, 174, 175 133, 147, 148, 161 GPCR, xxi, 41, 50, 104, 106, 182–184, human gut microbiome, 122 http://www.sol 187–190, 192, 193, 196, 203–221, 223, Human Microbiome Project, 119, 130, 224, 226, 264, 275, 283 143, 156 host on on host arrestin, 101, 102, 189, 194, 204, hybrid molecules, 41 206, 208 hyperconjugation, 94 arrestin-based assay, 219 arrestin-GPCR, 203 induced ﬁt, 215, 216, 220, 221 arrestin signaling, 206 inﬂammation, 130, 131, 155, 156, 254

http://www.worldscientific.com/worldscibooks/10.1142/ bacteriorhodopsin, 179, 186, 226 institutes DREADDs, 226 Academia Sinica, 165 G protein, 180, 183, 186, 189, 190, DOE Joint Genomics Institute, 24 192, 196, 204, 206, 208, 211, Federal Drug Administration, 160, 213–216, 221, 222, 264 161, 231, 233, 271, 272, 274 G protein-biased agonist, 206 National Institute of Health, 24, 25, G protein-coupled receptor, 101, 47, 75, 119, 168, 216, 273, 276 180, 186, 195 integral membrane proteins, 182

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World Gαβγ complex, 213 intein, 42 G-CSF receptor, 270 interferon, 231, 244, 272 NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY GPCR-ome, 197, 219 intracellular proteolysis, 204 multidrug transport, 210, 223 iodothymidine, 54 opsin, 215, 223 iodotyrosine, 54

rhodopsin, 115, 186, 188, 205, 215, isoprenaline, 215 220–223 isotope eﬀect, 65–68, 94, 96 green ﬂuorescent protein, 104 growth hormone, 42, 231, 272 lactose, 175 October 14, 2014 16:35 WSPC Proceedings - 9.75in x 6.5in master/pg 352

352 Index

lanthionine, 7 174, 175 last eukaryotic common ancestor microbiota, xxi, 119, 120, 130, 131, 133, (LECA), 147, 148 144, 146, 157, 158 Le Chatelier, 215, 216 microﬂuidics, 47, 252

leptin, 269 microswitch, 204

lipocalin, 234, 243–245 molecular mechanics, 100, 308 lithium, 214 Morse curves, 101 luminal content, 155 mRNA, 4, 240, 308–312, 315, 316, 327 multi-domain assembly, 36 macromolecular machine, 344 muscarinic, 216 mass spectrometry, 38, 45, 81, 98, 270, mutagenesis, 11, 21, 59, 60, 65–67, 84, 274 104, 219, 245 distribution is allowed. distribution HCD fragmentation, 166 mutation, 7, 19, 48, 49, 60, 66, 70, 72,

membrane permeability, 237, 238, 241 82, 84, 86–88, 102, 104–108, 112, 113, membrane proteins, 151, 179, 180, 184, 116, 158, 210, 212, 213, 221, 260, 262, 9180 187, 192, 193, 199, 210, 225–227, 234, 320, 337, 340, 341 250, 267, 273, 274, 324, 329, 333, 335, mutational drifts, 53 only. No further No only.

343 mycoplasma, 155, 161 7-transmembrane superfamily, 186, 188–190 N-acetylglucosamine, 150 ABC exporter, 199, 200, 222, 223 native chemical ligation, 11, 13, 126, 239

vayinstitutes.be/ ABC importer, 199–201 natural product, 6, 7, 26, 33, 38, 40, 44,

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS ABC transporter, 199–201, 208, 46, 47, 53, 78, 103, 161, 241, 299, 300,

- 210, 222, 223 303, 338 GPCR, see also GPCR New York Structural Genomics lipidic cubic phase, 179, 209 Research Consortium, 76 http://www.sol potassium channel, 44 NFκB, 155 receptors, see also receptors N-hydroxysuccinimide, 168 host on on host vitamin B12 transporter BtuCD-F, niacin, 205, 206 201 N-methylated amino acid, 241, 280 mesenchymal, 283 noradrenaline, 102 messenger RNA, 80, 308, 326 norfenﬂuramine, 195 metabolic network, 34, 120, 122 nucleophilic, 54, 64, 106

http://www.worldscientific.com/worldscibooks/10.1142/ metabolite, 34, 75, 76, 99, 119, 122, 123, 134, 147, 155, 161, 195, 243 open-reading frame, 43, 54 metabolomics, 26, 75, 77, 156 optogenetics, 213 metadata, 162, 163 orphan receptors, 196 meta-genome, 19, 24, 32, 33 OST complex, 225 metagenomics, 23, 25, 120, 123, 130, oxazole, 4, 277 131, 133, 142–144, 146, 148, 162 oxygen-sensitive, 35, 54, 55 metallation, 91 oxytocin, 42

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World metallo-biochemistry, 35 metallo-enzyme, 39 para-nitrophenol, 97 NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY meta-proteomics, 43 parasite, 138, 161, 262, 263, 267, 315 methyllanthionine, 7 peptides microbial protein, 36, 123, 131 acylheptapeptide, 44

microbiologist, 33, 38 bicyclic, 13, 14, 40, 240 microbiome, xxi, 23–25, 75, 103, bioactive, 52, 238 117–123, 130–134, 136, 141–144, cyclic peptide, 9, 13, 40, 42, 79, 146–152, 154, 156, 157, 161, 163, 168, 238, 241 October 23, 2014 9:30 WSPC Proceedings - 9.75in x 6.5in master/pg 353

Index 353

isopeptide bond, 40 phosphorylation, 7, 21, 37, 41, 46, 150, lanthipeptides, 7 204 lasso-peptide, 40 photosynthesis, 49 lipoheptapeptide, 44 poly-β-ketone, 300, 301, 340

microcyclic peptide, 280 polycyclic compound, 40, 53

non-ribosomal peptide synthesis, 3, poly-ester, 340 55, 299, 300, 303, 339 polyethylene glycol, 244 oligopeptides, 136 polyketide, 301, 338 polycyclic, 9, 11, 13, 300 polyketide synthase, 299, 300, 302, 338, ribosomally synthesized and 339 post-translationally modiﬁed polyketide synthesis, 55 peptides, 6–9 polyprotein, 323 distribution is allowed. distribution solid-phase synthesis, 11–14, 137, post-translational amidation, 44

163, 173 post-translational modiﬁcation, 3, 4, 6, peptidyl-tRNA, 305, 309, 336 7, 20–22, 37, 45, 46, 53, 81, 205, 224, 9180 periplasm, 99, 183, 288, 289, 340 225, 233, 238, 241, 242, 250, 270, 271, perivascular, 254, 268, 284 306 only. No further No only. Peyer’s patches, 161 prion protein, 138, 171 pharmaceutical companies, 6, 29, 42, probiotic, 156, 162 159, 211, 272, 279 prokaryotic transporter, 210 Amylin Pharmaceuticals, 44 prostaglandin, 102

vayinstitutes.be/ Codexis, 72, 106 protein structure

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS Dendreon, 281 active site, 16, 17, 19, 32, 48, 49,

- DuPont, 109 60, 61, 64–68, 71, 72, 76, 82, 85, Genencor, 110 90, 97–99, 102, 105, 111, 112, Genentech, 159, 169, 213, 266, 269, 116, 132, 225, 237, 300, 302, 303, http://www.sol 284 311, 321, 338, 340 GlaxoSmithKline, 277 ancestral fold, 156 host on on host Heptares, 218 bromodomains, 29 Immunogen, 278 crystallography, 34, 67, 208, 209, Ironwood Pharmaceuticals, 42 295, 308, 309, 314, 318, 324 Novozymes, 110 dynamic motion, 68, 100, 150, 314, Pﬁzer, 30, 163, 206, 209, 279 339

http://www.worldscientific.com/worldscibooks/10.1142/ Receptos, 218 four-helix bundle, 226 Roche, 270 GPI anchor, 137–139, 171 Roche-Trimeris, 43 hydrogen-bond network, 204 Seattle Genetics, 277, 278 hydrophobic core, 50 Shell Gas, 110 hypervariable loop, 169, 243, 244, TransCure, 201, 210 324 pharmaceuticals, 33, 42, 99, 103, 104, lectin domain, 287, 290, 291, 326, 126, 159, 160, 182, 231, 244, 247, 252, 340

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World 256 nucleotide-binding domain, 200 pharmaco-chaperones, 212 orthosteric binding site, 182, 183, NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY pharmacokinetics, 159, 237, 238, 269 216 pharmacology, 104, 195, 197, 209, 213 protein dynamics, 48, 102, 303 phenotype, 41, 134, 162, 169, 257, 258, protein fold, 16, 36, 141, 244

269, 270, 274, 281, 306 secondary structure, 13, 50, 52, 305 phenotypic screen, 9, 274 α-helix, 335 phosphonic acid tail, 168 β-barrel, 50, 243, 244, 289, 327 phosphoryl transfer, 68, 113 Rossmann fold, 17, 32, 328 October 14, 2014 14:31 WSPC Proceedings - 9.75in x 6.5in master/pg 354

354 Index

TIM barrel, 32, 33, 116 318, 320, 326, 327, 329, 331, 332, type-II β-turn, 164 334–336, 338, 341, 343 SH2 domain, 250, 251 RNAi, 32, 38, 232, 263 structural scaﬀold, 32, 37, 49, 50 rubidium, 214

substrate-binding pocket, 64 RuBisCo, 18, 48, 49

protein universe, xix, xxi, 3, 4, 6, 9, 11, 14, 16, 23–26, 55, 62, 119, 147 Saccharomyces cerevisiae, 81, 151, 169 proteomics, xxi, 4, 26, 43, 130–132, 137, S-adenosylmethionine, 67, 91, 94, 95 156 scientists protium, 65, 66, 95, 96 Anderson, Vernon, 95 pseudouridylated tRNA, 332 Baker, David, 52, 61, 70, 71, 97 psychrophile, 111 Belfort, Marlene, 341 distribution is allowed. distribution pyrophosphate, 98 Budisa, Nediljko, 98

pyrrolidine, 305, 336 Bukau, Bernd, 334 Cahill, Dolores, 274 9180 radical SAM, 54, 55, 91 Danishefsky, Samuel, 160, 164 random non-standard peptides Dennis, Mark, 269 only. No further No only. integrated discovery, 240, 241 Donohue, John Paul, 308 rare codon, 327, 328 Ehrlich, Paul, 243, 247 reactive oxygen, 90 Fischbach, Michael, 161 receptors, 45, 120–123, 183, 186–188, Frankel, Alan, 331

vayinstitutes.be/ 195, 204, 207, 209–211, 213, 214, Frasier, Claire, 37

PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS 216–221, 224, 225, 235, 266, 279, 325, Furnham, Nick, 17

- 337 Ghose, Sraboni, 231 β receptor, 222 Guo, Zhuojun, 308 β1 adrenergic receptor, 102, 214 Hartl, Ulrich, 334 http://www.sol β3, 217 Hofmann, Peter, 223 5hT2A-mGluR2, 217 Jacobson,Matthew,99 host on on host adenosine A2A receptor, 187, 221 Kobilka, Brian, 186, 187, 190, 194, adrenergic receptor, 186, 192 213, 216 amine receptor, 220 Kornberg, Roger, 32 angiotensin receptor, 206 Lancaster, Laura, 308 apo-β2AR, 212 Lefkowitz, Robert, 206

http://www.worldscientific.com/worldscibooks/10.1142/ asialoglycoprotein receptor, 279 Lehrach, Hans, 274 chemokine receptor, 220 Lesage, Anne, 205 C receptor, 216 Martinez, Todd, 94, 100 dopamine D3, 189, 219 Mayford, Mark, 226 Fcγ receptor, 284 McCaﬀerty, John, 266 muscarinic receptor, 220, 226 Natsume, Tohru, 275 nociceptin receptor, 189, 217 Nelson, Karen, 154 nuclear hormone receptor, 34 Nichols, Jeremy, 154

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World opioid receptor, 189, 214, 217 Pardon, Els, 192 serotonergic, 196 Puglisi, Joseph, 331 NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY serotonin receptor, 189, 195, 209 Ramakrishnan, Venki, 332 vasopressin receptor, 212 Roberts, Ed, 218 β2AR, 193, 194 Rosen, Hugh, 218

reproducibility, 271 Rudinger, Josef, 42 retrograde transport, 165, 167 Schwyzer, Robert, 42 ribosome, 45, 79, 80, 91, 109, 180, 225, Sherman, David, 339 239, 280, 305, 306, 308–312, 314, 315, Shoichet, Brian, 196, 219, 220 October 14, 2014 16:39 WSPC Proceedings - 9.75in x 6.5in master/pg 355

Index 355

Steinman, Ralph, 161 tRNA, 4, 79, 80, 109, 239, 286, 305, Tate, Chris, 210 308–312, 315, 316, 326, 327, 333, 341 Thiel, Walter, 100 Trypanosoma brucei, 286, 315 Trypanosomes, 262–264, 267, 334

Thornton, Janet, 17 von Behring, Emil, 243, 247 tubulin, 330 Ward, Andrew, 324 Ward, Thomas, 111 ubiquitin, 40, 81, 235, 251, 276 Weissman, Jonathan, 45 ubiquitination, 37, 204 Wintermeyer, Wolfgang, 335 unnatural amino acids, 21, 39, 79–82, Yusupov, Marat, 332 91, 98, 107, 108, 331 Zhou, Jie, 308 3-methyllanthionine, 7 dehydroalanine, 4, 7

distribution is allowed. distribution SEC translocon, 224 dehydrobutyrine, 7 secretion, 225, 329 nonproteinogenic, 3 selective pressure, 87, 102 unnatural ligand, 104

9180 sequence analysis, 50, 53, 301 serotonin, 209 van der Waals, 66, 67, 95, 101 only. No further No only. Shine-Dalgarno-like sequence, 328 virus, 138, 171, 174, 257, 259, 293, 336 short-chain fatty acids, 134 viruses, 131, 146, 175, 267, 295, 314 sialic acid, 159, 294 baculovirus, 169 signal transduction, 62, 127, 203, 270 Ebola, 294 signaling molecules, 33, 193, 206 HIV, see also HIV vayinstitutes.be/ starch, 26, 141, 175 inﬂuenza, 126, 127, 138, 169, 267, PROCEEDINGS OF THE 23RD INTERNATIONAL SOLVAY CONFERENCE ON ON CHEMISTRY CONFERENCE SOLVAY INTERNATIONAL THE 23RD OF PROCEEDINGS

- statins, 206 293–296, 337 stereospeciﬁcity, 159 lentivirus, 258, 269, 270 steroid, 102–104 phage, 157, 235, 245, 250, 257 http://www.sol structural genomics, xxi, 4, 16, 25, 119, Rhinovirus, 258 120 viral antigen, 169, 171, 295

host on on host Structural Genomics Consortium, 29, 30, 187 X-ray crystallography, 120, 182, 193, sucrose, 175 316 sumoylation, 21 surfactin, 44 yeast, 36, 41, 47, 80, 150, 160, 166, 179,

http://www.worldscientific.com/worldscibooks/10.1142/ synthetic biology, 7–9, 47, 59 301, 302

Tat-TAR interaction, 331 taxonomicaldrift,172 thermostabilisation, 99 thiazole, 4, 277 three-dimensional structure, 16, 17, 29, 51, 156, 193, 233, 291, 313, 331

©World Scientific Publishing Company. For For Company. Publishing Scientific ©World tomogram, 330 toxin, 84, 138, 231, 247, 261, 277, 284, NEW CHEMISTRY AND NEW OPPORTUNITIES FROM THE EXPANDING PROTEIN UNIVERSE UNIVERSE PROTEIN EXPANDING THE FROM OPPORTUNITIES NEW AND NEW CHEMISTRY 300, 301 trace metals, 38 traceless isotopes, 97

transcriptional regulation, 205 transcriptomics, 75, 77, 156 translocon, 180, 225, 329, 333, 335 tritium, 67, 94, 95